CENTER FOR NUTRITION POLICY AND PROMOTION Family Economics ... · PDF file2 Family Economics...

Research Articles 3 Expenditures on Children by Families Mark Lino

11 Consumers' Retail Source of Food: A Cluster Analysis Andrea Carlson, Jean Kinsey, and Carmel Nadav

21 Food, Health, and Supplements: Beliefs Among Food Stamp-Eligible Women Vivica Kraak, David L. Pelletier, and Jamie Dollahite

36 Factors Affecting Meat Preferences Among American Consumers Arbindra P. Rimal

44 Food Acquisition Practices Used by Limited-Resource Individuals Kathryn M. Kempson, Debra Palmer Keenan, Puneeta Sonya Sadani, Sylvia Ridlen, and Nancy Scotto Rosato

56 Trends in Food and Nutrient Intakes by Children in the United States Cecilia Wilkinson Enns, Sharon J. Mickle, and Joseph D. Goldman

69 Trends in Children's Consumption of Beverages: 1987 to 1998 Yi Kyung Park, Emily R. Meier, Peri Bianchi, and Won O. Song

Research Briefs80 Insight 23: The Role of Nuts in a Healthy Diet Mark Lino, Kristin Marcoe, Julia M. Dinkins, Hazel Hiza, and Rajen Anand

83 Insight 24: Food Trade-Offs: Choosing How to Balance the Diet Julia M. Dinkins and Mark Lino

Regular Items

USDA Activities Federal Studies Journal Abstracts Food Plans Consumer Prices Poverty Thresholds

Family Economics and Nutrition Review

CENTER FOR NUTRITION POLICY AND PROMOTION

Volume 14, Number 22002

.. .. .

Ann M. Veneman, SecretaryU.S. Department of Agriculture

Eric M. Bost, Under SecretaryFood, Nutrition, and Consumer Services

Steven N. Christensen, Acting DirectorCenter for Nutrition Policy and Promotion

P. Peter Basiotis, DirectorNutrition Policy and Analysis Staff

The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs andactivities on the basis of race, color, national origin, gender, religion, age, disability, politicalbeliefs, sexual orientation, or marital or family status. (Not all prohibited bases apply to allprograms.) Persons with disabilities who require alternative means for communication ofprogram information (Braille, large print, audiotape, etc.) should contact USDA’s TARGETCenter at (202) 720-2600 (voice and TDD).

To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room326-W, Whitten Building, 14th and Independence Avenue, SW, Washington, DC 20250-9410 or call (202) 720-5964 (voice and TDD). USDA is an equal opportunity provider andemployer.

Family Economics andNutrition Review

Research Articles

3 Expenditures on Children by FamiliesMark Lino

11 Consumers’ Retail Source of Food: A Cluster AnalysisAndrea Carlson, Jean Kinsey, and Carmel Nadav

21 Food, Health, and Nutrient Supplements: Beliefs Among Food Stamp-EligibleWomen and Implications for Food Stamp Policy

Vivica Kraak, David L. Pelletier, and Jamie Dollahite

36 Factors Affecting Meat Preferences Among American ConsumersArbindra P. Rimal

44 Educators’ Reports of Food Acquisition Practices Used by Limited-ResourceIndividuals to Maintain Food Sufficiency

Kathryn M. Kempson, Debra Palmer Keenan, Puneeta Sonya Sadani, Sylvia Ridlen,and Nancy Scotto Rosato

56 Trends in Food and Nutrient Intakes by Children in the United StatesCecilia Wilkinson Enns, Sharon J. Mickle, and Joseph D. Goldman

69 Trends in Children’s Consumption of Beverages: 1987 to 1998Yi Kyung Park, Emily R. Meier, Peri Bianchi, and Won O. Song

Research Briefs

80 Insight 23: The Role of Nuts in a Healthy DietMark Lino, Kristin Marcoe, Julia M. Dinkins, Hazel Hiza, and Rajen Anand

83 Insight 24: Food Trade-Offs: Choosing How to Balance the DietJulia M. Dinkins and Mark Lino

Regular Items

86 Research and Evaluation Activities in USDA

90 Federal Studies

98 Journal Abstracts

100 Official USDA Food Plans: Cost of Food at Home at Four Levels, U.S. Average,December 2002

101 Consumer Prices

102 U.S. Poverty Thresholds and Related Statistics

103 Reviewers of 2002 Articles

Volume 14, Number 22002

EditorJulia M. Dinkins

Assistant EditorDavid M. Herring

Features EditorMark Lino

Managing EditorJane W. Fleming

Peer Review CoordinatorHazel Hiza

Family Economics and Nutrition Review iswritten and published semiannually by theCenter for Nutrition Policy and Promotion, U.S.Department of Agriculture, Washington, DC.

The Secretary of Agriculture has determined thatpublication of this periodical is necessary in thetransaction of the public business required bylaw of the Department.

This publication is not copyrighted. Contentsmay be reprinted without permission, but creditto Family Economics and Nutrition Reviewwould be appreciated. Use of commercialor trade names does not imply approval orconstitute endorsement by USDA. FamilyEconomics and Nutrition Review is indexed inthe following databases: AGRICOLA, Ageline,Economic Literature Index, ERIC, FamilyStudies, PAIS, and Sociological Abstracts.

Family Economics and Nutrition Review is forsale by the Superintendent of Documents.Subscription price is $13 per year ($18.20 forforeign addresses). Send subscription orderand change of address to Superintendent ofDocuments, P.O. Box 371954, Pittsburgh, PA15250-7954. (See subscription form on p. 104.)

Original manuscripts are accepted forpublication. (See “guidelines for authors” onback inside cover.) Suggestions or commentsconcerning this publication should be addressedto Julia M. Dinkins, Editor, Family Economicsand Nutrition Review, Center for Nutrition Policyand Promotion, USDA, 3101 Park Center Drive,Room 1034, Alexandria, VA 22302-1594.

The Family Economics and NutritionReview is now available at http://www.cnpp.usda.gov. (See p. 2.)

CENTER FOR NUTRITION POLICY AND PROMOTION

2 Family Economics and Nutrition Review

Check the CNPP Web site (www.cnpp.usda.gov)for the following information and publications:

• The Healthy Eating Index: 1999-2000• “Get on the Grain Train”• “How Much Are You Eating?”• Dietary Guidelines for Americans, 2000, 5 th Edition• Interactive Healthy Eating Index• Recipes and Tips for Healthy, Thrifty Meals• About CNPP• CNPP Strategic Plan 2000-05• How to Get Information from CNPP• Nutrition Insights• Dietary Guidelines for Americans• Food Guide Pyramid• Food Guide Pyramid for Young Children• USDA Healthy Eating Index• Expenditures on Children by Families• Family Economics and Nutrition Review• Nutrient Content of the U.S. Food Supply Summary Report• Interactive Nutrient Content of the U.S. Food Supply• Official USDA Food Plans• Putting the Guidelines into Practice: A Series of Brochures• Web-Based Training on “The ABCs of the Dietary Guidelines 2000:

Science and Application”• USDA Supports 5 A Day for Better Health• Making Healthy Food Choices• Nutritional Status of WIC Participants Study• Miscellaneous Files• Symposia Proceedings• Video Archives in Real Video

United States Department ofAgriculture

CENTER FOR NUTRITIONPOLICY AND PROMOTION

2002 Vol. 14 No. 2 3

Expenditures on Childrenby Families

Since 1960 the U.S. Department of Agriculture has provided annual estimates ofexpenditures on children from their birth through age 17. This article presents the2001 estimates for husband-wife and single-parent families. Data and methodsused in calculating annual child-rearing expenses are described. Estimates areprovided by budgetary component, age of the child, family income, and region ofresidence. For the overall United States, estimates of child-rearing expensesranged between $9,030 and $10,140 for a child in a two-child, married-couplefamily in the middle-income group.

hild rearing is a costly endeavor.Since 1960 the U.S. Departmentof Agriculture (USDA) has

provided annual estimates of familyexpenditures on children from theirbirth through age 17. USDA’s annualchild-rearing expense estimates areused in four major ways:

• To determine State child supportguidelines. Under the FamilySupport Act of 1988, States arerequired to have numeric childsupport guidelines and to considerthe economic costs of raising a childin these guidelines. The economicwell-being of millions of childrenis affected by child support.

• To determine State foster carepayments. In 1999 about 581,000children were in foster care (U.S.Department of Health and HumanServices, 2001).

• To appraise damages arising from personal injury or wrongful deathcases. For example, if a person withchildren is hurt on a job such that heor she cannot work, the courts usethe expense figures to determinecompensation for the family.

• To educate anyone consideringwhen or whether to have children.These expense estimates also mayencourage teens to wait until theyare adults and more preparedfinancially to have children.

This article presents the 2001 expendi-ture estimates associated with rearingchildren. Data and methods used incalculating the child-rearing expensesare described; then, the estimatedexpenses are discussed.

USDA Method for EstimatingExpenditures on Children byFamilies1

USDA provides annual estimates ofexpenditures on children, by husband-wife and single-parent families, fromtheir birth through age 17. Expendi-tures on children are estimated for the

1Expenditures on Children by Families: 2001Annual Report provides a more detaileddescription of the data and methods. To obtaina copy go to http://www.cnpp.usda.gov, orcontact USDA, Center for Nutrition Policyand Promotion, 3101 Park Center Drive,Room 1034, Alexandria, VA 22302(telephone: 703-305-7600).

Mark Lino, PhDU.S. Department of AgricultureCenter for Nutrition Policy and Promotion

C


major budgetary components: housing,food, transportation, clothing, healthcare, child care/education, andmiscellaneous goods and services(see box below).

The most recently calculated child-rearing expenses are based on 1990-92Consumer Expenditure Survey (CE)data, which are updated to 2001 dollarsby using the Consumer Price Index(CPI). The CE, administered by theBureau of Labor Statistics, U.S.Department of Labor, is the onlyFederal survey of household expendi-tures collected nationwide. It containsinformation on sociodemographiccharacteristics, income, and expendi-tures of a nationally representativesample of households. The sampleconsisted of 12,850 husband-wifeand 3,395 single-parent households,weighted to reflect the U.S. populationof interest.

In determining child-rearing expenses,USDA examines the intrahouseholddistribution of expenditures by usingdata for each budgetary component.The CE contains child-specificexpenditure data for some budgetarycomponents (clothing, child care, andeducation) and household-level datafor the other budgetary components(housing, food, transportation, healthcare, and miscellaneous goods andservices). Multivariate analysis wasused to estimate household and child-specific expenditures, controlling forincome level, family size, age of thechild, and region of residence (whenappropriate) so expenses could bedetermined for families with thesevarying characteristics. Estimates ofchild-rearing expenses are provided forthree income levels of husband-wifefamilies. These income groups weredetermined by dividing the sample forthe overall United States into equalthirds.

For each income level, the estimatesare for the younger child in familieswith two children. These youngerchildren were grouped in one of sixage categories: 0-2, 3-5, 6-8, 9-11,12-14, or 15-17. Households with twochildren were selected as the standardbecause this was the average householdsize in 1990-92. The focus is on theyounger child because the older childmay be over age 17.

USDA’s estimates are based on CEinterviews of households with andwithout specific expenses. For somefamilies, expenditures may be higheror lower than the mean estimates,depending on whether or not they incurthe expense. Child care and educationare examples, since about 50 percentof husband-wife families in the studyspent no money on these services.Also, the estimates cover only out-of-pocket expenditures on children madeby the parents and not by others, suchas grandparents or friends.

Categories of Household Expenditures

Housing expenses: shelter (mortgage interest, property taxes, or rent; maintenance and repairs; and insurance), utilities(gas, electricity, fuel, telephone, and water), and house furnishings and equipment (furniture, floor coverings, and majorand small appliances). For homeowners, housing expenses do not include mortgage principal payments; in the data setused, such payments are considered to be part of savings.

Food expenses: food and nonalcoholic beverages purchased at grocery, convenience, and specialty stores, includingpurchases with food stamps; dining at restaurants; and household expenditures on school meals.

Transportation expenses: the net outlay on the purchase of new and used vehicles, vehicle finance charges, gasoline andmotor oil, maintenance and repairs, insurance, and public transportation.

Clothing expenses: children’s apparel such as diapers, shirts, pants, dresses, and suits; footwear; and clothing servicessuch as dry cleaning, alterations and repair, and storage.

Health care expenses: medical and dental services not covered by insurance, prescription drugs and medical supplies notcovered by insurance, and health insurance premiums not paid by the employer or other organizations.

Child care and education expenses: daycare tuition and supplies; babysitting; and elementary and high school tuition,books, and supplies.

Miscellaneous expenses: personal care items, entertainment, and reading materials.

2002 Vol. 14 No. 2 5

After estimating the various overallhousehold and child-specific expendi-tures, USDA allocated these totalamounts among family members (i.e.,in a married-couple, two-child family,the total amounts were allocated to thehusband, wife, older child, and youngerchild). Because the expenditures forclothing, child care, and education arechild-specific and thus apply only tochildren, allocations of these expenseswere made by dividing them equallyamong the children. The CE doesnot collect expenditures on food andhealth care. Thus, to apportion thesebudgetary components to a child by hisor her age, USDA used data from otherFederal studies that show the shares ofthe household budget spent onchildren’s food and health care.

Unlike food and health care, noauthoritative source exists forallocating among family membersthe amount the household spendson housing, transportation, and othermiscellaneous goods and services. Twocommon approaches used to allocatethese expenses are the marginal costmethod and the per capita method.

The marginal cost method measuresexpenditures on children as thedifference in expenses between coupleswith children and equivalent childlesscouples. Various equivalency measureshave been proposed, yielding verydifferent estimates of expenditureson children, with no standard measureaccepted by economists. Also, themarginal cost approach assumes thatthe difference in total expendituresbetween couples with and withoutchildren can be attributed solely to thepresence of children in a family. Thisassumption is questionable, especiallybecause couples without childrenoften buy homes larger than theyneed in anticipation of having children.Comparing the expenditures of thesecouples to those of similar couples withchildren could lead to underestimating

how much is spent on meeting thelifetime needs—and wants—ofchildren.

For these reasons, USDA uses theper capita method to allocate expenseson housing, transportation, and miscel-laneous goods and services amonghousehold members in equal propor-tions. Although the per capita methodhas its limitations, these limitationsare considered less severe than thoseof the marginal cost approach. Becausetransportation expenses resulting fromwork activities are not directly relatedto the cost of raising a child, theseexpenses were excluded when deter-mining children’s transportationexpenses.

Expenditures on Children byHusband-Wife Families

Child-Rearing Expenses andHousehold Income Are PositivelyAssociatedEstimated expenses on childrenincreased as income level rose (fig. 1).Depending on the age of the child, theannual expenses ranged from $6,490to $7,560 for families in the lowestincome group, from $9,030 to $10,140for families in the middle-incomegroup, and from $13,410 to $14,670for families in the highest incomegroup. The before-tax income in 2001for the lowest income group was lessthan $39,100, between $39,100 and$65,800 for the middle-income group,and more than $65,800 for the highestincome group.

On average, households in the lowestincome group spent 28 percent of theirbefore-tax income per year on a child;those in the middle-income group, 18percent; and those in the highest group,14 percent. The range in these percent-ages would be narrower if after-taxincome were considered, because a

Estimated expenses on childrenincreased as income level rose(fig. 1).


0-2 3-5 6-8 9-11 12-14 15-17

Age of child

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

Dol

lars

Less than $39,100 $39,100 - $65,800 More than $65,800

Figure 1. 2001 family expenditures on a child, by income level and age ofchild1

1U.S. average for the younger child in husband-wife families with two children.

greater percentage of income in higherincome households goes toward taxes.

On average, the amount spent onchildren by families in the highestincome group was slightly less thantwice the amount spent by families inthe lowest income group. This amountvaried by budgetary component. Ingeneral, expenses on a child forgoods and services considered tobe necessities (e.g., food and clothing)did not vary as much as thoseconsidered to be discretionary (e.g.,miscellaneous expenses) amonghouseholds in the three income groups.

Housing Is the Largest Expenseon a ChildHousing accounted for the largest shareof total child-rearing expenses; figure 2demonstrates this for middle-incomefamilies. Based on an average expenseincurred among the six age groups,housing accounted for 33 percent ofchild-rearing expenses for a child in thelowest income group, 35 percent in themiddle-income group, and 38 percentin the highest income group. Food, thesecond largest average expense on achild for families regardless of incomelevel, accounted for 20 percent ofchild-rearing expenses in the lowestincome group, 17 percent in themiddle-income group, and 15 percentin the highest income group. Trans-portation was the third largest child-rearing expense across income levels,13 to 14 percent.

Across the three income groups,miscellaneous goods and services(personal care items, entertainment,and reading materials) was the fourthlargest expense on a child for families,10 to 12 percent. Clothing (excludinggifts or hand-me-downs) accounted for5 to 7 percent of expenses on a childfor families; child care and education,8 to 11 percent; and health care, 6 to 8percent. Estimated expenditures forhealth care included only out-of-

Figure 2. 2001 family expenditure shares on a child from birth through age 171

1U.S. average for the younger child in middle-income, husband-wife families with two children.

Total expenditures in 2001 dollars = $170,460

Food

Housing

Miscellaneous

Child care and education

Health care

Clothing

Transportation

35%

17%

11%

10%

7%

6%

14%

2002 Vol. 14 No. 2 7

pocket expenses (including insurancepremiums not paid by an employeror other organizations) and not thatportion covered by health insurance.

Expenses Increase as aChild AgesExpenditures on a child were generallylower in the younger age categoriesand higher in the older age categories.Figure 3 depicts this for families in themiddle-income group. This relationshipheld across income groups even thoughhousing expenses, the highest child-rearing expenditure, generally declinedas a child grew older. The decline inhousing expenses reflects diminishinginterest paid by homeowners over thelife of a mortgage. Payments onprincipal are not considered part ofhousing costs in the CE; they aredeemed to be a part of savings.

For all three income groups, food,transportation, clothing, and healthcare expenses related to child-rearinggenerally increased as the child grewolder. Transportation expenses were

highest for a child age 15-17, when heor she would start driving. Child careand education expenses were highestfor a child under age 6. Most of thisexpense may be attributed to child careat this age. The estimated expense forchild care and education may seemlow for those with the expenses:these estimates reflect the averageof households with and without theexpense.

Child-Rearing Expenses AreHighest in the Urban WestChild-rearing expenses in the regionsof the country reflect patterns observedin the overall United States; in eachregion, expenses on a child increasedwith household income level and,generally, with the age of the child.Overall, child-rearing expenses werehighest in the urban West, followed bythe urban Northeast and urban South.Figure 4 shows total child-rearingexpenses by region and age of a childfor middle-income families. Child-rearing expenses were lowest in theurban Midwest and rural areas. Much

0-2 3-5 6-8 9-11 12-14 15-17

Age of child

0

20

40

60

80

100

Per

cent

Miscellaneous

Child care and educationClothingHealth care

Transportation

Food

Housing

$9,030 $9,260 $9,260 $9,190 $9,940 $10,140

Figure 3. 2001 family expenditure shares on a child, by age of child1

1U.S. average for the younger child in middle-income, husband-wife families with two children.

Expenditures on a child weregenerally lower in the youngerage categories and higher in theolder age categories.


of the difference in expenses on a childamong regions was related to housingcosts. Total housing expenses on achild were highest in the urban Westand urban Northeast and lowest inrural areas. However, child-rearingtransportation expenses were highestfor families in rural areas. This likelyreflects the longer traveling distancesand the lack of public transportation inthese areas.

Children Are“Cheaper by the Dozen”The expense estimates on a childrepresent expenditures on the youngerchild at various ages in a husband-wifehousehold with two children. It cannotbe assumed that expenses on the olderchild are the same at these variousages. The method for estimatingexpenses on the younger child wasessentially repeated to determinewhether expenses vary by birth order.The focus was on the older child ineach of the same age categories asthose used with the younger child.A two-child family was again usedas the standard.

On average, for husband-wifehouseholds with two children,expenditures did not vary by birthorder. Thus, annual expenditures onchildren in a husband-wife, two-childfamily may be estimated by summingthe expenses for the two appropriateage categories reported in figure 1.

Although expenses on children didnot vary by birth order, they did differwhen a household had only one childor more than two children. Dependingon the number of other children in thehousehold, families spent more or lesson a child—achieving a “cheaper-by-the-dozen” effect as they have morechildren. That is, the cost of twochildren is less than double thecost of one child.

The method to estimate child-rearingexpenses was repeated for familieswith one child and families with threeor more children. Compared withexpenditures for each child in ahusband-wife family with two children,husband-wife households with onechild spent an average of 24 percentmore on the single child; those withthree or more children spent an averageof 23 percent less on each child.Hence, family income is spread overfewer or more children, subject toeconomies of scale. As families havemore children, the children can share abedroom, clothing and toys can behanded down to younger children, andfood can be purchased in larger andmore economical packages.

Expenditures on Children bySingle-Parent Families

The estimates of expenditures onchildren by husband-wife families do

0-2 3-5 6-8 9-11 12-14 15-17

Age of child

8,000

8,500

9,000

9,500

10,000

10,500

11,000

11,500

12,000

Dol

lars

Urban MidwestRural

Urban SouthUrban Northeast

Urban West

Figure 4. 2001 family expenditures on a child, by region and age of child1

1Regional averages for the younger child in middle-income, husband-wife families with two children.

not apply to single-parent families, agroup that accounts for an increasingpercentage of families with children.Therefore, separate estimates weremade of child-rearing expenses insingle-parent households for the overallUnited States. CE data were used to doso. Most single-parent families in thesurvey were headed by a woman (90percent). The method previouslydescribed was followed; regionalestimates were not calculated forsingle-parent families because oflimitations in the sample size.

Estimates cover only out-of-pocketchild-rearing expenditures made by thesingle parent with primary care of thechild and do not include child-relatedexpenditures made by the parentwithout primary care or made byothers, such as grandparents. The datadid not contain this information.Overall expenses by both parents on achild in a single-parent household arelikely greater than the USDA child-rearing expense estimates.

2002 Vol. 14 No. 2 9

Table 1 presents estimated expendi-tures on the younger child in a single-parent family with two children,compared with those of the youngerchild in a husband-wife family withtwo children. Each family type was inthe lower income group, having before-tax income less than $39,100. About83 percent of single-parent familiesand 33 percent of husband-wifefamilies were in this lower incomegroup. More single-parent thanhusband-wife families, however, werein the bottom range of this incomegroup, and had an average incomeof $16,400, compared with $24,400for husband-wife families. Althoughaverage income varied for thesefamilies, total expenditures on a childthrough age 17 were, on average, only5 percent lower in single-parent house-holds than in two-parent households.

Single-parent families in this lowerincome group, therefore, spent a largerproportion of their income on childrenthan did their counterpart two-parentfamilies. On average, housing expenseswere higher for single-parent familiesthan for two-parent families, whereastransportation, health care, child careand education, and miscellaneousexpenditures on a child were lower

in single-parent than in husband-wifehouseholds. Child-related food andclothing expenditures were similar,on average, for both family types.

For the higher income group of single-parent families with 2001 before-taxincome of $39,100 and over,2 estimatesof child-rearing expenses were aboutthe same as those for two-parent house-holds in the before-tax income group of$65,800 and over. In 2001 dollars, totalexpenses for the younger child throughage 17 were $250,260 for single-parentfamilies versus $249,180 for husband-wife families. Child-rearing expensesfor the higher income group of single-parent families, therefore, were also alarger proportion of income than wasthe case for husband-wife families.Thus, expenditures on children do notdiffer much between single-parent andhusband-wife households; what differsis household income levels. Becausesingle-parent families have one lesspotential earner than do husband-wifefamilies, on average, their total house-hold income is lower, and child-rearingexpenses are a greater percentage ofincome.

2The two higher income groups were combinedfor single-parent families.

The same procedure was used toestimate child-rearing expenses on anolder child in single-parent householdsas well as by household size. Onaverage, single-parent households withtwo children spent 7 percent less on theolder child than on the younger child(in addition to age-related differences).This contrasts with husband-wifehouseholds whose expenditures onchildren were unaffected by thechildren’s birth order.

As with husband-wife households,single-parent households spent moreor less if there was either one childor there were three or more children.Compared with expenditures for theyounger child in a single-parenthousehold with two children, expendi-tures for an only child in a single-parent household averaged 35 percentmore; households with three or morechildren averaged 28 percent less oneach child.

Other Expenditures onChildren

The USDA child-rearing expenseestimates consist of direct expensesmade by parents on children throughage 17 for seven major budgetarycomponents. The expenses excludecosts related to childbirth and prenatalhealth care and other expenditures,especially those incurred after a childturns age 18.

One of the largest expenses made onchildren after age 17 is the cost of acollege education. The College Boardestimated that in 2001-2002, annualaverage tuition and fees were $3,586 at4-year public colleges and $14,456 at4-year private colleges; annual roomand board was $4,956 at 4-year publiccolleges and $5,704 at 4-year privatecolleges (The College Board, 2001).Other parental expenses on children

Table 1. 2001 family expenditures on a child, by lower income single-parentand husband-wife households1

Single-parent Husband-wifeAge of child households households

0 - 2 $5,440 $6,490 3 - 5 6,150 6,630 6 - 8 6,910 6,710 9 - 11 6,440 6,730 12 - 14 6,920 7,560 15 - 17 7,670 7,480Total (0 - 17) $118,590 $124,800

1Estimates are for the younger child in two-child families in the overall United States.


after age 17 could include thoseassociated with children living at homeor, if children do not live at home, giftsand other contributions to them. A1996 survey found that 47 percent ofparents in their fifties support childrenover age 21 (Phoenix Home LifeMutual Insurance Company, 1996).

USDA’s estimates do not include allgovernment expenditures on children,such as public education, Medicaid,and subsidized school meals. Actualexpenditures on children (by parentsand the government), therefore, wouldbe higher than reported here. Theindirect costs of raising children—time allocated to child rearing anddecreased earnings—are not includedin the estimates. Although these costsare more difficult to measure thandirect expenditures, they can be just ashigh, if not higher, than the direct costsof raising children (Spalter-Roth &Hartmann, 1990; Bryant, Zick, & Kim,1992; Ireland & Ward, 1995).

Estimated annual expenditures on a child born in 2001, by income group, overallUnited States1

1Estimates are for the younger child in husband-wife families with two children.2Total reflects expenses on a child through age 7.

The estimates presented in this articlerepresent household expenditures ona child of a certain age in 2001. Futureprice changes need to be incorporated toestimate these expenses over time. Thus,a future cost formula was used, and theresults are presented in the graph below.The estimated future expenditures areon the younger child in a husband-wifefamily with two children. The assump-tions are that a child is born in 2001and reaches age 17 in 2018 and that theaverage annual inflation rate over thistime is 3.4 percent (the average annualinflation rate over the past 20 years). Theresult: total family expenses on a childthrough age 17 would be $169,920 forhouseholds in the lowest income group,$231,470 for those in the middle, and$337,690 for those in the highest incomegroup.

2001

2005

2010

2015

2018

Year

$0

$5

$10

$15

$20

$25

$30

Tho

usan

ds

Lowest Middle Highest

Total = $337,690

Total = $169,920

Total = $231,470

References

Bryant, W.K., Zick, C.D., & Kim, H. (1992). The Dollar Value of HouseholdWork. College of Human Ecology, Cornell University, Ithaca, NY.

The College Board. (2001). Trends in College Pricing 2001. Retrieved April 2,2002, from http://www.collegeboard.org.

Ireland, T.R., & Ward, J.O. (1995). Valuing Children in Litigation: Family andIndividual Loss Assessment. Tucson, AZ: Lawyers and Judges PublishingCompany, Inc.

Lino, M. (2002). Expenditures on Children by Families: 2001 Annual Report.U.S. Department of Agriculture, Center for Nutrition Policy and Promotion.Miscellaneous Publication No. 1528-2001. Available at www.cnpp.usda.gov.

Phoenix Home Life Mutual Insurance Company. (1996). Americans’ Hopes,Fears and Dreams. 1996 Phoenix Fiscal Fitness Survey.

Spalter-Roth, R.M., & Hartmann, H.I. (1990). Unnecessary Losses: Costs toAmericans of the Lack of Family and Medical Leave. Washington, DC: Institutefor Women’s Policy Research.

U.S. Department of Health and Human Services, Administration on Children,Youth, and Families, Children’s Bureau. (2001). The AFCARS Report. RetrievedApril 25, 2002, from http://www.acf.dhhs.gov/programs/cb/publication/afcars/june2001.pdf.

Child-Rearing Expenses Over Time

2

2002 Vol. 14 No. 2 11

Consumers’ Retail Source of Food:A Cluster Analysis

The popular impression that only half of our food comes from retail grocerystores is based on food expenditure data. However, the U.S. Department ofAgriculture’s Continuing Survey of Food Intakes by Individuals, 1994 showsthat 72 percent of the amount of food (measured in grams) consumed byAmericans comes from grocery stores. Using cluster analysis, we groupedconsumers based on where they obtained their food and found that half were“Home Cookers”—purchasing 93 percent of their food from grocery stores. Bycomparison, the “High Service” consumers, which represented 10 percent ofthe sample, purchased 43 percent of their food from restaurants. This researchquantifies the different shopping behaviors exhibited by groups of people in theUnited States and discusses some of the demographic differences among theclusters. The results are of interest to consumers, nutrition counselors, foodretailers, and policymakers who deal with retail food, low-income diets, or foodsafety.

Andrea Carlson, PhDU.S. Department of AgricultureCenter for Nutrition Policy and Promotion

Jean Kinsey, PhDUniversity of Minnesota

Carmel Nadav, PhDVenturi Technology Partners

n economic analysis of consumerbehavior, substituting expenditurefor quantity is a common practice.

For example, expenditure is often sub-stituted for quantity when estimatingthe percentage change in the amountconsumed when income changes by1 percent (Engel function). This sub-stitution is often used because expendi-ture data rather than quantity are morefrequently available. And from abusiness perspective, expendituresare more closely related to sales—theindicator (or metric) most used bybusinesses to measure demand for theirproducts. Tracking consumers’ foodconsumption behavior with expendituredata is no exception: the percentageof income spent on food is a commonmeasure of economic well-being bothfor individual households and fornations.

The percentage of personal disposableincome spent on food by Americanconsumers decreased from 25 to 11

percent between 1960 and 1997 (Putnam& Allshouse, 1996). The composition ofthose expenditures changed noticeably,with a decreasing proportion of eachfood dollar being spent on food from aretail food store called “food at home.”Food-away-from-home expenditures,according to the food service andrestaurant sector, grew from 26 to 45percent of each food dollar between1960 and 1994; by the end of 1995, theamount reached 47 percent (Putnam& Allshouse, 1996). In recent years,expenditures on food away from homehave approached 50 percent (Putnam& Allshouse, 1996).

The rapid rise in food-away-from-homeexpenditures is reflected in anothermetric: the high growth in sales atcommercial food service establishmentsrelative to the growth in sales in retailfood stores. Between 1987 and 1999,inflation-adjusted sales in eating anddrinking establishments grew anaverage of 2.2 percent; similar sales in

I


retail food stores, however, decreasedan average 0.1 percent (Food Institute,1997).

Focusing on the proportion of the fooddollar that is spent in places other thana grocery store leads to the commonbelief that Americans eat almost halfof their food away from home. Theamounts of food consumers eat at homeor away from home, however, variesconsiderably from the expenditureproportions reported in the literature.Expenditures in food service establish-ments reflect higher costs of labor(about 30 percent of the menu price),entertainment, and service.

In contrast, we reported in 1998 thatwhen food consumption is measured ingrams, the amount of food purchasedfrom retail stores is 72 percent of allfood consumed (Carlson, Kinsey, &Nadav, 1998). Another 14 percent offood (in grams) was consumed fromcarryout establishments (e.g., fast-food,pizza, and sandwich shops) and otherrestaurants combined. The remaining14 percent came from other sources—other people and gifts, cafeterias,vending machines, coffee or food ona common tray in an office, bars andtaverns, home gardens or hunting andfishing, and public programs. Whenfood consumption is measured byexpenditure, the amount of food (g)consumed away from home is 47percent, almost twice as much asthat consumed from restaurants,carryouts, and other establishments.

Our earlier research also found thatwhere people purchase their food didnot necessarily predict where theyconsumed their food. For example,10 percent of food purchased in storeswas not consumed at home, while 24percent of carryout food was consumedat home (Carlson, Kinsey, & Nadav,1998). Rising household incomes andfewer hours for household labor foretella rising value of time and, in turn,

predict that consumers will purchasemore labor services in their pursuit offood (Kinsey, 1983). Even within agrocery store, sales of ready-to-eatfoods—including those that must beheated—are rising while sales for basicingredients are falling.

Studies in the 1970’s and 1980’s foundthat higher incomes led consumers tospend more money on meals eaten outbut did not necessarily lead consumersto eat more meals away from home(Prochaska & Shrimper, 1973). A similarconclusion from other research sug-gests that households with wives whowork part-time increased their expendi-tures on food away from home moreso than did households where wivesworked full-time even though bothhouseholds had the same income(Kinsey, 1983).

As women’s time in the labor marketexpands from zero to part-time,increases in income may expand theopportunity to eat out. But as employ-ment becomes full-time, less time isavailable to eat out or cook at home.Thus, continued increases in incomeare not further associated with in-creased expenditures on food awayfrom home. In fact, increases in incomemay even decrease expenditures onfood away from home as consumerssubstitute fast-foods or take-out foodsfor more leisurely dining away fromhome (Kinsey, 1983). These findingssuggest that the traditional labels of“food at home” and “food away fromhome,” as well as the use of expenditureas the metric for quantity, do notprovide a complete understandingof today’s consumer.

The research reported here investigatesthe amount of food (g) that consumersreported eating in 1994 from variousretail sources and examines the commoncharacteristics of consumers whoseretail sources of food vary from theaverage. We used data from the USDA

Continuing Survey of Food Intakes byIndividuals, 1994 (CSFII) (USDA, 1994).We examined two questions: (1) Whatare the unique characteristics of peoplewho shop for food in different typesof establishments? (2) How can thisinformation be used by managers ofthese establishments and publicpolicymakers? To answer thesequestions, we used cluster analysis togroup consumers by the retail source oftheir food and to describe their commonshopping and eating habits.

Data and Methods

The CSFII is conducted by the Agricul-tural Research Service (USDA, 1994).1

We used data from 1994 because theywere the most recent data availablewhen this study began. The CSFIIdata provide a better picture of overallconsumption behavior than do datacollected at the market level wheresales are the unit of measure. TheCSFII reports all food eaten by 5,589individuals in 2,540 households in theUnited States. Each individual reportsfood intake for 2 nonconsecutive days,yielding more than 150,000 observa-tions on individual food items. Forevery food item, the respondent alsolists the source from which the foodwas obtained and how much waseaten. The sources of food used inthis analysis include stores, carryoutrestaurants, restaurants, other people,bars and taverns, cafeterias, commoncoffee pots or trays, vending machines,mail order, public programs, and home-grown or caught food (see box). Theresponse rate for the CSFII is 80 percentfor the first day and 76 percent for thesecond day. Sample weights are usedin this analysis, and the results aregeneralizable to the population.

1These data are available from the U.S.Department of Commerce, TechnologyAdministration, National Technical Service,5285 Port Royal Road, Springfield, VA22161, (703) 487-4650, http://www.ntis.gov.

2002 Vol. 14 No. 2 13

Analysis

The first step in our analysis wasto calculate the percentage of food,measured in grams, each personconsumed from each source. Clusteranalysis is used to place the adultsample2 into groups based on wherethey obtained their food. In this case,the cluster variables are the percentageof food (g) adults consumed that comefrom various sources. For example, ifone person’s diet contains 80 percentof food from stores, 5 percent fromcarryout restaurants, 10 percent fromrestaurants, and the remaining 5 percentfrom cafeterias, cluster analysis usesthese percentages to place that personinto a group with others who havesimilar consumption patterns.

This analysis uses the “k-means”method of clustering that is used bySAS FASTCLUS. This method is oneof the better techniques available forclustering large data sets where thegoal is to divide respondents intomanageable and meaningful groups todescribe behavior (Hartigan, 1985; SASInstitute, 1989). K-means selects thecenters of the initial clusters from thefirst observations in the data set andthen assigns the other observations tothe nearest cluster. When an observa-tion is added to the cluster, k-meansrecalculates the mean of the clustervariables, and this mean becomes thenew cluster-center. If this recalculatedcluster-center changes another clusterthat is closest to an observation alreadyin the cluster, then k-means moves thatobservation to the closest cluster andrecalculates the center of its newcluster. The process continues untilthe number of changes is very small.

2Because children’s eating behaviors aresomewhat dictated by their parents, childrenare not included in the cluster analysis.

Categories of Food Sources

Store: supermarket, grocery store, warehouse, convenience store, drug store,gas station, bakery, deli, seafood shop, ethnic food store, health food store,commissary, produce stand, and farmers’ market.

Carryout: traditional hamburger, chicken, and carryout pizza restaurants; andother restaurants where customers order, pick up, and pay for food at a counter.

Restaurants: any other establishment where the food is served at the table byrestaurant staff.

Other People: food received as a gift or while a guest in someone’s home.

Bars and Taverns: a location the respondent classified as a bar or tavern ratherthan as a restaurant, carryout restaurant, or cafeteria.

School and Non-School Cafeterias: Most non-school cafeterias are basedin offices. For most of the analysis, school and non-school cafeterias areseparated but are often put together in summary tables.

Common Coffee Pot or Food Tray: office coffee pots, food platters at areception or in an office, and potluck dinners.

Vending Machines: food purchased from vending machines located withinstores, restaurants, cafeterias, offices, or other locations.

Mail Order: food received from a mail order catalog or club that sends food outregularly, such as a fruit-of-the-month club.

Public Programs: a combination of several CSFII categories including childand adult care centers, day care centers in private homes, soup kitchens,shelters, food pantries, Meals on Wheels, other community food programs,and residential care facilities.

Home-Grown or Caught: food that is grown or gathered by the respondent orsomeone the respondent knows; meat and fish procured by hunting or fishing.


The resulting clusters are based on 2nonconsecutive days of dietary recall.Thus, if an individual had been sampledon a different day, he or she might haveended up in a different cluster. How-ever, because this data set is designedto be nationally representative, similarclusters would form on any day, exceptmajor national holidays.

To reduce the bias towards observa-tions that appear at the beginning ofthe data set, we used a techniquerecommended by SAS (SAS Institute,1989). In the first pass, the SASprocedure forms 50 clusters and savesthe cluster centers in a file. Over halfof these clusters have fewer than fiveobservations, and the centers areignored. The remaining 24 centers formthe “seeds” in the next iteration to form24 new clusters. In the third iteration,the center of the smallest cluster isremoved, and the SAS procedure forms23 new clusters from all observations.This process continues until there arefive clusters. The process is describedin more detail elsewhere (Carlson,Kinsey, & Nadav, 1998; MacQueen,1967).

The second step compared each clusterwith the rest of the sample to addressthe two research questions. Becausemost of the data were categorical, thisstudy used three nonparametric tests:the chi-squared, the Kolmogrov-Smirnov test, and the Kruskal-Wallistest (described in detail elsewhere)(Siegel, 1956). These tests measuredifferences in distributions of variablesamong different subgroups. The chi-squared test was used as an initial testfor differences. Differences between theobserved versus expected distributionswere confirmed by the other two tests.The Kolmogrov-Smirnov test was usedto measure differences between twoclusters in the distribution of categori-cal variables that cannot be ranked(e.g., race) and the Kruskal-Wallis testfor differences in categories that can be

ranked (e.g., age, income, and educa-tion). For these tests, we divided thecontinuous variables into categories.For example, the categories for agewere 19-30, 31-40, 40-50, 50-60, 60-64,and 65+; for education, less than highschool, high school degree or GED,some college, 4-year degree, andprofessional or graduate study.

Results and Discussion

Nineteen clusters formed around thevarious sources of food. Severalsources, such as carryout, had morethan one cluster form around it. Thispaper will discuss only nine of theseclusters, some with names based on theunique characteristics of the cluster:Working Family, Young Professional,Manager, and City Office. In othercases, the names are based on wherethe people in the cluster shopped:Home Cookers, Carryout, High Service,Office, and Students and Faculty.

SociodemographicCharacteristics of the SampleAlmost half (49 percent) of the adultsample was in the Home Cookerscluster (table 1), followed by those inthe Working Family cluster (11 percent),and High Service cluster (10 percent).Fewer adults were in the other clusters:Carryout, Office, Manager, YoungProfessional, City Office, and Studentsand Faculty (from 3 percent to a low of0.6 percent).

Age, Race, and Gender. With anaverage age of 51, people in the HomeCookers cluster were significantlyolder than the rest of the adult sample(tables 1 and 2). However, the standarddeviation for their age was the largest(17.9, not shown), indicating a biggerspread in age than was the case forthe other clusters. Three clusters—Students and Faculty, Carryout, andYoung Professional—had the youngestmembers (mean age of 37, 36, and 31,

Representing 75 percent of theadult sample, six of the nineclusters get more food fromstores than any other source . . .

2002 Vol. 14 No. 2 15

Table 1. Statistically significant demographic characteristics of select clusters of consumers based on where they purchasedfood

Percent of Age, race, Income and Occupation and Region, urban, andCluster adults1 and gender education employment household size

Home Cookers 49 Older** Lower income** Fewer professional/Less college** technical, and

manager/proprietor**More not employed**

Working Family 11 Younger** More “some college”* More full- and part-time** Larger households**

Carryout 3 Younger than More full- and part-time* More Northeast**Working Family**Fewer White*

Young Professional 0.7 Younger than Higher income** More full-time**Carryout** More college and

graduate study**

High Service 10 More White** Higher income** More professional/More men* More college** technical, and

manager/proprietor*More full-time**

Office 2.5 More full-time**

Manager 2.0 Higher income** More professional/ More central city*More college/university** technical, and

manager/proprietor*More full-time**

City Office 1.0 More full-time** More central city*

Students and Faculty 0.6 More Asian/Pacific More college and More full- and part-time* More Northeast**and “other”** graduate**Fewer females*

1Percents do not add to 100, because all clusters are not shown in the table.* p<.05; ** p<.01: The distribution between the cluster and the rest of the adult sample is significantly different based on the Kruskal-Wallis test.

respectively). Whereas significantlymore Whites were in the High Servicecluster, fewer Whites were in theCarryout cluster, and more Asian/Pacific Islanders and others were inthe Students and Faculty cluster.

The High Service cluster had signifi-cantly fewer women (46 percent),3

3Differences are in the distributions betweenthe cluster and the total adult sample. The p-values do not indicate how these distributionsdiffer, only that they are different.

compared with the remainder of theadult sample. The Young Professionalcluster also had relatively few women(35 percent), but the difference from theadult sample was not significant. TheYoung Professional cluster, however,represented only 0.7 percent of thetotal sample; thus, the small size ofthis cluster may have contributed tothe lack of statistical significance.

Income, Education, and Employment.Mean income among the clustersranged from $32,554 to $49,072.Compared with the rest of the sample,the Home Cookers cluster had asignificantly lower income; threeclusters had a higher income: HighService ($42,767), Young Professional($48,507), and Manager ($49,072).Although people in the WorkingFamily and Carryout clusters earneda household income close to theHome Cookers’ income ($36,466 and


$34,555, respectively), the distributionof incomes in the Working Familyand Carryout clusters did not differsignificantly from the rest of thesample.

Educational patterns tended to followincome patterns. Whereas the HomeCookers cluster had a significantlylower educational level, compared withthe total sample, several other clustershad higher levels of education: YoungProfessional, Students and Faculty,High Service, Manager, and WorkingFamily. The Young Professional andStudents and Faculty cluster each hadmore people with 4-year collegedegrees and graduate or professionaldegrees. About 83 percent each ofthe members of the Working Family,Carryout, City Office, and Managerclusters graduated from high schoolor received more education. Of these,only the Manager cluster, with moremembers receiving college anduniversity degrees, had a distributionthat was significantly different fromthe sample. Although not significantlydifferent from the rest of the sample,

76 percent of those in the Office clusterhad a high school degree or more.

Occupation and Employment. The HomeCookers cluster, compared with theHigh Service and Manager clusters,had significantly fewer people inprofessional/technical occupations orwho worked as managers/proprietors.Compared with other clusters, the HomeCookers cluster was significantly morelikely to have unemployed members—and a concentration of unemployedpeople (including retirees). Whereasonly 47 percent of the people in theHome Cookers cluster were employed,most of the people in the YoungProfessional cluster were employed(96 percent). A little more than three-fourths of those in the Manager clusterwere employed (77 percent).

Region, Urbanization, and HouseholdSize. Two clusters, Carryout as well asStudents and Faculty, were more likelythan other clusters to reside in theNortheast. Two clusters, Manager andCity Office, had a higher percentageof people living in center cities, 46 and

52 percent, respectively. Household4

size among all the clusters rangedfrom an average of 2.7 to 3.4. Onlythe distribution for the Working Familycluster differed significantly from therest of the sample. The Carryout andYoung Professional also appeared tohave larger households (3.2 and 3.4,respectively), but the distributionswere similar to the remainder of theadult sample.

Food SourcesRepresenting 75 percent of the adultsample, six of the nine clusters get morefood from stores than any other source:Home Cookers (93 percent), Office (73percent), Working Family (70 percent),Students and Faculty (54 percent),Manager (53 percent), and High Service(47 percent) (table 3).

When using grams of food rather thanexpenditure as a measure of consumerbuying behavior, we found that stores

4This analysis did not include children, but wedid examine the number of children present inthe households of the adult respodents.

Table 2. Basic sociodemographic characteristics of select clusters of consumers based on where they purchased food

High school Adult Center city degree HouseholdCluster sample Women resident or more Employed size Age Income

----------------------------------------- Percent ---------------------------------------- ----------------------- Mean ------------------------------Entire Adult Sample 100 49.8 33.3 76.6 57.5 2.9 48.3 $35,298Home Cookers 49.0 51.4 33.4 71.0 46.5 2.9 51.4 32,554Working Family 10.0 47.8 31.4 82.7 65.7 3.2 41.8 36,466Carryout 11.0 45.2 40.9 82.6 78.3 3.2 36.0 34,555Young Professional 3.0 34.8 30.4 91.3 95.7 3.4 30.8 48,507High Service 0.7 45.5 35.1 85.6 62.1 2.8 48.3 42,767Office 2.4 55.7 39.2 76.0 73.4 3.0 49.0 39,824Manager1 2.0 45.6 45.6 82.5 77.2 2.7 46.8 49,072City Office 0.7 52.2 52.2 82.6 91.3 2.8 41.5 35,963Students and Faculty 1.0 68.8 21.9 90.6 87.5 3.2 36.8 44,361

1Includes a high concentration of professionals, technical workers, managers, and proprietors.

2002 Vol. 14 No. 2 17

appear to play a much more importantrole for most consumers. A secondobservation is that both carryoutrestaurants and cafeterias have morethan one cluster purchasing foods (g)from them, indicating major differencesbetween the customers using thesepoint-of-purchase sources. Threeclusters formed around carryout food:Working Family, Carryout, and YoungProfessional. There are also differencesin the shopping patterns, especiallyin the amount of food obtained fromcarryout restaurants, 22 to 57 percent.In addition, the Young Professionalcluster is the only cluster discussedin this paper with a relatively high useof vending machines (14 percent).Similarly, four clusters formed aroundcafeterias as a source of food. TheOffice, Manager, and City Officeclusters formed around non-schoolcafeterias, while the Students andFaculty cluster formed around schoolcafeterias (breakdown not shown).Except for City Office, these clustersall get at least half of the remainingfood from stores, and make use ofrestaurants and carryout restaurants,though in different proportions.

Market ProfilesWhen we examined consumption withinmarkets (e.g., stores), we found thatHome Cookers, the largest cluster,consumed 59 percent of all food (g)obtained from stores (fig. 1). The nexttwo biggest clusters, Working Familyand High Service, consumed 10 and6 percent, respectively, of all foodobtained from this source. This patternof larger clusters representing largerportions of this market continued.“Other Groups” are clusters thatformed but are not discussed in thispaper. Each of these clusters in “OtherGroups” had fewer than 100 observa-tions; thus, statistical analysis maybe misleading.

For restaurants, carryout restaurants,and cafeterias, the largest market sharebelonged to the cluster or clusterswhich formed around that source. Forexample, the High Service cluster, whichformed around restaurants, represented58 percent of the restaurant’s marketshare. For carryout restaurants, theWorking Family, Carryout, and YoungProfessional clusters consumed overthree-fifths (61 percent) of all foodobtained from that market. Whereas

Table 3. The percentage share of food source for select clusters of consumersbased on where they purchased food

Cluster Food store Restaurant Carryout Vending Cafeteria1

Home Cookers 93.1 2.5 1.1 0.3 0.1Working Family 69.6 3.3 22.0 0.3 0.2Carryout 34.8 3.7 57.3 0.1 0.2Young Professional 33.8 8.2 40.4 14.2 0.8High Service 46.8 42.8 5.0 0.4 0.4Office 72.6 4.2 3.8 0.7 14.7Manager 52.7 7.3 4.3 1.0 28.1City Office 27.9 7.0 7.3 2.6 52.8Students and Faculty 54.2 8.3 6.8 1.1 25.0

1Both school and non-school cafeterias are combined.Notes: Bold numbers identify the behavior around which a cluster was formed.Totals do not add to 100, because not all sources of food are shown. . . . Home Cookers, the largest

cluster, consumed 59 percentof all food (g) obtained fromstores.


Home Cookers (59%)

Working Family (10%)

Carryout (2%)Office (3%)High Service (6%)Manager (1%)

Other Groups (18%)

Students & Faculty (1%)

Figure 1. Percentage of food consumed from selected sources, by cluster

Home Cookers (20%)

Working Family (6%)Carryout (2%)Young Professional (1%)

High Service (58%)

Manager (2%)

Other Groups (8%)


Office (2%)

Home Cookers (13%)

Working Family (34%)

Carryout (23%)

Office (1%)High Service (7%)Manager (1%)

Other Groups (15%)


Young Professional (4%)

City Office (1%)

Home Cookers (4%)Working Family (1%)Carryout (0.5%)Office (12%)

High Service (2%)

Manager (17%)

Other Groups (10%)

Students & Faculty (42.5%)

City Office (11%)

Store Restaurant

Carryout Cafeteria

Note: “Other groups” are clusters with fewer than 100 observations; these clusters are not discussed in this paper but are needed to complete the market profile.

2002 Vol. 14 No. 2 19

the High Service cluster consumed 7percent of the food in this market, theYoung Professional cluster consumedless, 4 percent. However, the HighService cluster is a much larger cluster.

For the Carryout market, 70 percent ofall food obtained here was consumedby three clusters: Working Family(34 percent of the grams of food con-sumed), Carryout (23 percent), andHome Cookers (13 percent). Asexpected, the Students and Faculty,Managers, Office, and City Officeclusters consumed 83 percent of thefood in the school and non-schoolcafeteria market. No other clusterconsumes a large part of their foodfrom this source, indicating thecafeteria market is fairly focusedon these four clusters.

Conclusion

Americans who report in detail whatfood they eat, where they eat it, andwhere they buy it provide us with analternative picture of food consumptionbased on the quantity of food (g)consumed. This varies from the morecommon picture based on food expendi-tures and sales. While it is true thatAmericans obtain food from many retailand home-grown sources, 75 percentof the adult population purchasedover half of their food measured ingrams from retail food stores. Thuswe have a very different picture fromthe one presented by the use of foodexpenditure data. This alternativepicture allowed us to ask twoquestions, what are the uniquecharacteristics of people who shopfor food in different establishments,and how can this information be usedby these establishments and by publicpolicymakers?

An examination of the data to determinethe importance of each cluster to eachtype of retail vendor shows that,

among the people in our sample, HomeCookers purchase 59 percent of all thegrams of food that were sold in retailstores, 20 percent of restaurant food,and 13 percent of the food fromcarryout establishments. The clustersmost likely to be consumers of carryoutfood were the Young Professional,Working Family, and Carryout. Peoplein these groups tend to be younger,employed, and have some collegeeducation.

Policymakers can use this informationto determine how policies will affectdifferent market segments: stores,restaurants, cafeterias, or carryoutestablishments. Owners and marketersof these establishments can determinewhere else their customers are obtain-ing food and design an appropriatemarketing strategy.

Future research needs to addressthe effect that the choice of where toobtain food has on the quality andhealthfulness of the diet. Identifyingthe consumers who are the first to makechanges to their shopping habits, aswell as identifying their preferences,will help retailers and those who designpublic food policy to serve consumersbetter.


References

Carlson, A., Kinsey, J., & Nadav, C. (1998). Who Eats What, When and FromWhere? Minneapolis, MN: The Retail Food Industry Center, University ofMinnesota. Working Paper Series.

Food Institute. (1997). U.S. Food Service Industry Segments . Food InstituteReview, 44, 3.

Hartigan, J.A. (1985). Statistical Theory in Clustering. Journal of Classification,2, 63-76.

Kinsey, J. (1983). Working wives and the marginal propensity to consume foodaway from home. American Journal of Agricultural Economics, 65, 10-19.

MacQueen, J.R. (1967). Some Methods for Classification and Analysis ofMultivariate Observations. Paper presented at the Fifth Berkeley Symposium onMathematical Statistics and Probability, Berkeley, CA.

Prochaska, F., & Shrimper, R. (1973). Opportunity cost of time and othersocioeconomic effects on away-from-home-food consumption. American Journalof Agricultural Economics, 66, 595-603.

Putnam, J.J., & Allshouse, J.E. (1996). Food Consumption, Prices, andExpenditures, 1970-94. U.S. Department of Agriculture, Economic ResearchService. Statistical Bulletin No. 928.

SAS Institute, Inc. (1989). SAS/STAT User’s Guide, Version 6 (4th ed.). Cary, NC:SAS Institute, Inc.

Siegel, S. (1956). Nonparametric Statistics for the Behavioral Sciences. New York:McGraw-Hill.

U.S. Department of Agriculture, Agricultural Research Service. (1994). ContinuingSurvey of Food Intakes by Individuals (CSFII) .

2002 Vol. 14 No. 2 21

Food, Health, and NutrientSupplements: Beliefs AmongFood Stamp-Eligible Women andImplications for Food Stamp Policy

Several U.S. professional organizations that develop research-based dietaryrecommendations for the public support the position that most nutrients can andshould be obtained by consuming a balanced diet. This position differs from thewidespread and growing use of supplements by the public and changes in publicpolicy currently under consideration, such as the proposal to allow nutrientsupplements to be purchased with food stamps. This study investigated theattitudes and beliefs of a diverse sample of food stamp-eligible women concerningthe relationship among food, health, nutrient supplementation, and associatedlifestyle factors; these findings were then related to ongoing policy dialogue. Thefindings suggest the need to clarify the policy goals, conduct a more systematicexamination of potential strategies for achieving those goals, and broaden theset of explicit criteria used when considering supplement-related policies in thispopulation.

Vivica Kraak, MS, RDDavid L. Pelletier, PhDJamie Dollahite, PhD, RD

Cornell University

any U.S. organizations thatdevelop research-basednational dietary recommenda-

tions support the position that nutrientsrequired by healthy people can beobtained by consuming a balanceddiet (Pelletier & Kendall, 1997). TheAmerican Dietetic Association main-tains that “the best nutrition strategy forpromoting optimal health and reducingthe risk of chronic disease is to obtainadequate nutrients from a wide varietyof foods” (Hunt, 1996). The FoodGuide Pyramid and the DietaryGuidelines for Americans, 2000 alsosupport this perspective by promotinga food-based approach for U.S.consumers to achieve optimal health(Johnson & Kennedy, 2000). The useof supplements,1 however, is a growingtrend, which suggests that Americansare becoming more receptive to non-food sources of nutrition for healthpromotion.

A recent biannual nationwide surveyconducted by the American DieteticAssociation (2002), which trackspublic attitudes, beliefs, knowledge,and practices related to food, nutrition,and health, found that nearly half (49percent) of the adults surveyed tooksupplements daily, and more than athird (38 percent) believed that takingsupplements is necessary to ensure

1Nutrient supplements are defined by the U.S.Department of Health and Human Services(DHHS), Office of Dietary Supplements as aformulation containing at least one or moreof a variety of vitamins and minerals used tosupplement the diet by increasing the totaldietary intake. Dietary supplements, abroader class of products, include a vitamin,mineral, amino acid, herb, or other botanicalintended for ingestion in the form of a capsule,powder, soft gel, or gel cap, and which is notrepresented as a conventional food or as a soleitem of a meal in the diet (Office of DietarySupplements, 1999).

M


good health. The high prevalence ofsupplement use has been confirmed inother national surveys (Balluz, Kieszak,Philen, & Mulinare, 2000; Bender,Levy, Schucker, & Yetley, 1992;Slesinski, Subar, & Kahle, 1995;Subar & Block, 1990). Further analysessuggest that users of nutrient supple-ments tend to have higher incomes andeducation and more healthful lifestylesthan do nonusers (Nayga & Reed,1999; Neuhouser, Patterson, & Levy,1999), although supplement use alsois associated with having one or morehealth problems (Bender et al., 1992;Newman et al., 1998). Many studieshave reported that vitamin and mineralintakes from food tend to be higheramong supplement users than nonusers,but analysis of data from the 1989-91Continuing Survey of Food Intakes byIndividuals revealed that this relation-ship can vary across sociodemographicgroups and is influenced by themotivations and beliefs for usingsupplements (Pelletier & Kendall,1997).

In 1995 and 1999, Congress consideredlegislation to permit food stamprecipients to use their benefits topurchase dietary supplements (H.R.104-236 and S.1307, respectively)(Thomas, 2000). This legislation wasopposed by many organizations thatmonitor public health and hunger,including the American Academy ofPediatrics; American Heart Associa-tion; USDA; and the Food, Research,and Action Center (Pelletier & Kendall,1997; Porter, 1995; Skolnick, 1995).These organizations voiced a range ofconcerns: Most important was that apolicy change would depart from theoriginal intent of the Food StampProgram (FSP), and supplements wouldnot provide the calories needed or fullrange of nutritional benefits by childrento avoid health problems and maximizelearning potential in school (Porter,1995). The proposed change was alsoseen as an attempt by a billion-dollar

supplement industry to widen its market(Skolnick, 1995). USDA’s position onthis issue was stated in these terms:

Because vitamins and mineralsoccur naturally in foods, a gooddiet will include a variety of foodsthat together will supply all thenutrients needed. . . . Becausethese products serve as deficiencycorrectors or therapeutic agentsto supplement diets deficient inessential nutrition rather than asfoods, they are not eligible forpurchase with food coupons.(Porter, 1995)

Those favoring the proposed legislationmaintained that the bill would expanddietary choices by giving food stamprecipients the option of improving theirdiet through additional nutrients. TheCouncil for Responsible Nutrition, atrade organization representing thefood supplement industry, testified inCongress:

When critical food choicesare necessary, spending a fewcents a day for a vitamin andmineral supplement mayactually be the best and mosteconomical choice availableto a person at nutrition risk.(Dickinson, 1998)

Thus, supporters framed the issue interms of improving nutrition andmaintaining personal choice.

A report prepared by USDA at therequest of Congress examined issuesrelated to this proposal (U.S. Depart-ment of Agriculture [USDA], 1999).Among other findings, the report notedvitamin and mineral intake from fooddiffers little across income levels, foodstamp recipients tend to have nutrientprofiles that are comparable to non-recipients, and a third (35 percent) offood stamp recipients already purchasesupplements with other income sources.

The current policy, therefore, may notrestrict individual choice as some havesuggested.

There is a paucity of research elucidat-ing attitudes, beliefs, and supplement-use practices of low-income, ethnicallydiverse Americans. One study sug-gested that food stamp recipients areless likely to take dietary supplementsthan are nonrecipients. However, itanalyzed neither the reasons for thispractice nor the relationship to nutri-tional quality of the diets, health status,socioeconomic circumstances, or othercontextual factors (Nayga & Reed,1999). The purposes of the presentresearch were to investigate the atti-tudes and beliefs toward supplementuse among food stamp-eligible womento understand better the potentialeffects of policy changes in thispopulation and to relate these findingsto the earlier policy dialogue aboutthis issue, including the discussionof policy goals, strategies, and criteriafor selecting among them.

Methods

Study Sites and SamplingThe purpose of this research was toclarify the perspectives about nutrientsupplement use rather than to obtainpopulation-level estimates of thedistribution of particular beliefs.Qualitative methods were used bytwo researchers trained in qualitativeresearch techniques (Miles &Hubberman, 1994) to elucidateattitudes and beliefs of food stamp-eligible women concerning food,health, and nutrient supplements.Member checks and peer debriefing2

2A member check involves obtaining feedbackfrom respondents on the interpretation of thedata following the analysis; peer debriefinginvolves discussing the analysis and interpreta-tion of the data with other researchers (Miles &Hubberman, 1994).

2002 Vol. 14 No. 2 23

were techniques used by bothresearchers to enhance the reliabilityand credibility of the data (Kraak,Pelletier, & Dollahite, 2000).

Three study sites were selected toprovide ethnic and regional variationamong food stamp-eligible individualswho were nutrient supplement usersor nonusers. A purposeful sample wasobtained at each study site and wasbased on ethnicity (African American,White, Latina, and Asian), eligibilityfor food stamps (current recipientand/or former recipient), and use ofsupplements (user or nonuser). Eachcase was reviewed and classifiedaccording to the usual supplement-usehabits. For instance, women werecategorized as users if they occasion-ally used supplements whenever thesupplements were needed or when theyremembered to take them. By contrast,women were categorized as nonusersif they took a prenatal multivitamin/mineral only during pregnancy, asadvised by their physician, but did notuse supplements preceding or followingtheir pregnancy. The interviews wereconducted in urban locations includingNew York; San Francisco, San Jose,and Oakland, California; and FortSmith, Arkansas.

With the assistance of the CooperativeExtension staff in each site, we re-cruited 72 individuals—6 from eachethnic group in each location. Effortswere made to recruit participants whowere food stamp-eligible adult women,at least 18 years old, who had receivedor were receiving food stamps, andwere not pregnant or breastfeeding. Thefinal sample consisted of 24 individualsin New York (NY), 25 in California(CA), and 23 in Arkansas (AR).Participants in NY were drawn fromthe Expanded Food and NutritionEducation Program (EFNEP). Thosein CA and AR were drawn either fromthe EFNEP and Food Stamp NutritionEducation Program (FSNEP) or

contacted with the assistance oforganizations serving the populationthat met the sampling criteria.

The age range for the 72 participantswas 19 to 75 years. Thirty-eight of thefinal sample used supplements, 34 didnot; 37 were food stamp recipients, 34were not; and 1 respondent was unclearabout her use of supplements. Thefinal sample consisted of 19 Whites,16 African Americans, 20 Latinas,and 17 Asian Americans. Most inter-views were conducted in Englishamong bilingual interviewees; ininterviews with three Asian partici-pants, a bilingual interpreter was used.

Eligibility for EFNEP in the participat-ing States required a family incomeless than or equal to 185 percent of thepoverty level; whereas, eligibility forFSNEP was less than or equal to 130percent of the poverty level. Specificquestions about income were not asked,but participants were asked to identifyall of the food assistance programs theyknew they were eligible for and hadparticipated in. Some EFNEP partici-pants may have been ineligible toreceive food stamp benefits. Currentor former food stamp recipients madeup 38 percent of the sample in NY, 52percent in CA, and 65 percent in AR.

Interview Guide, Data CollectionMethods, and AnalysisQualitative methods were used fordata collection and analysis (Miles &Hubberman, 1994). A semi-structured,open-ended interview guide was used toelicit participants’ views and attitudesconcerning the following areas:

• attitudes about and participationin food assistance and nutritioneducation programs;

• eating habits;• beliefs about the adequacy of food-

based nutrients in the averageAmerican diet;

• beliefs about the general attributesof a healthy person;

• perceptions about their own healthstatus;

• personal health concerns;• health-promoting or health-

detracting behaviors;• intentions to adopt health-promoting

behaviors;• perceptions about the meaning of the

term supplement;• specific supplement-use habits;• influences promoting nutrient

supplement use;• reasons for not using or

discontinuing supplements;• beliefs about the benefits and

drawbacks of allowing the use offood stamps to purchase nutrientsupplements in addition to food; and

• opinions about who—thegovernment or food stamprecipients—should decide how foodstamps could be used if the policychanged.

After receiving input from staff of theCooperative Extension program, wepretested and modified the interviewguide for each site. Interviews weretaped and transcribed verbatim. Datafrom the transcripts, demographicinformation, and field notes were usedto analyze the qualitative data.

A consolidated summary was generatedfrom the ethnic- and geographicallybased summaries of pertinent emergentthemes. An in-depth analysis of keythemes was undertaken in four specificcategories:

• Nutrient supplement usersreceiving food stamp benefits

• Nutrient supplement users notreceiving food stamp benefits

• Nutrient supplement nonusersreceiving food stamp benefits

• Nutrient supplement nonusers notreceiving food stamp benefits


These categories emerged as moreimportant themes than the regionaland ethnic categories used to obtainthe heterogeneous sample. Emergentthemes were incorporated into aconceptual framework describing theattitudes, beliefs, and practices of thewomen. Examples were chosen toillustrate the breadth of results foreach question in the interview guide.The research site (NY, CA, AR),participants’ ethnicity, supplementstatus (user vs. nonuser), and foodstamp status (recipient vs. nonrecipient)are indicated after each quote. In somecases, approximate percentages areprovided to give a sense of the numberof women who expressed a certainviewpoint, although populationrepresentativeness should not beinferred.

Results

Attitudes and Beliefs About theAdequacy of Food-BasedNutrientsTwo major themes emerged from thisquestion: “Can the average person getall the vitamins and minerals he/sheneeds to be healthy, from the averageU.S. diet, without taking a multivitaminand mineral pill?”

Theme 1: Roughly 60 percent ofparticipants believe it is possible to getall nutrients from food, but most peopledo not do what is necessary to achievethat goal because of one or more of thefollowing:

• fast-paced and stressful lifestyles• ease and convenience of eating

“junk” food• lack of attention paid to the diet until

chronic diseases develop• lack of knowledge about what to

select and prepare to meet needs

• lack of precision in serving sizes toeat according to the Food GuidePyramid

• personal preferences that influencefood choices that may not benutritious

• time and money required to makewise decisions (especiallychallenging for low-income workingmothers)

• the perception that healthful foodsare too expensive to afford on alimited income

Theme 2: Less than one-quarter ofrespondents said it is not possible for aperson to obtain all necessary nutrientsexclusively from food because certainhealth conditions might require peopleto take nutrient supplements. Also,respondents had concerns about howfood is produced and processed withspecial reference to nutrient losses, useof pesticides, and food additives and/orpreservatives that were believed tochange the nutrient value of food.

Illustrations of the participants’attitudes and beliefs about the adequacyof food-based nutrients and the role ofnutrient supplements appear in the boxon page 25.

Perceptions Concerning theRole of a SupplementWhen participants were asked, “Whatcomes to mind when you hear the wordsupplement?” the responses followedthree themes. About one-half of theparticipants described a supplement asa substitute or a replacement for food.About one-quarter of the participantsdescribed it as something taken inaddition to the nutrients one couldobtain from food, and another quarterexpressed uncertainty about the purposeor role of a supplement.

About one-half of the participantsdescribed a supplement as asubstitute or a replacement forfood. About one-quarter of theparticipants described it assomething taken in addition tothe nutrients one could obtainfrom food, and another quarterexpressed uncertainty about thepurpose or role of a supplement.

2002 Vol. 14 No. 2 25

Illustrative statements of people’s attitudes and beliefs about adequacy of food-based nutrients and the roleof supplements

“Can the average person get all the vitamins and minerals he/she needs to be healthy from the average U.S. diet,without taking a multivitamin and mineral pill?”

Time and money to make wise food choices and/or to prepare nutritious foods were lacking.“Yes, if they eat right. If they have their diet balanced right, I believe they can, but most people don’t do that.It’s our culture . . . our society in America. You just slam food into your mouth and keep running. The way thegovernment has made it, people have to work to live, and they don’t take the time out for themselves. It’s really alabor of love. You really have to dedicate every day, commit, and I’m thinking about this right now . . . ‘How canI bring a lunch to work that’s more nutritious?’. . . A lot of people don’t have the time or energy to commit thatway.” CA, White, supplement user, former food stamp recipient

Food preferences influence food choices that may not provide all the nutrients people need.“No and that’s why I think I need to eat my vitamins because I don’t get enough. This food guide program of somuch of this and that. . . . I don’t follow it. I get three servings of fruits and vegetables out of five. . . . I am luckythat I like them. Even if I try, I am not very precise with my servings, and that is why I think I don’t get all thenutrients I need.” CA, Latina, supplement user, non-food stamp recipient

Food production techniques affect nutrient availability.“No, the good stuff [food] is too expensive because they [retailers] know it’s good. Sometimes when they grow itwith that . . . . I don’t know how to explain it, I don’t know all the terminology of it but for it to grow faster, itdoesn’t have all its nutrients, [and] half the time when people buy it, they don’t cook it right . . . . ” AR, White,supplement user, former food stamp recipient

Certain health conditions might require people to take supplements.“No . . . sometimes it is good to take vitamins . . . you go to the doctor and he prescribes for you how much youmust take and how often . . . some people need to take more because they don’t have enough of something orthey’ve become anemic . . . .” NY, White, supplement user, non-food stamp recipient

“What comes to mind when you hear the word supplement?”

A supplement is a substitute or replacement for food.“It’s like a second thing . . . that supplies . . . it’s a replacement I would think. . . . it’s like the fruits andvegetables; you can get better vitamins from them than pills, but sometimes when you can’t take all the foods thatyou need, you can take a pill . . . ; it’s not the best thing but it helps.” AR, Latina, supplement user, non-foodstamp recipient

A supplement adds extra to the nutrients obtained from food.“Something that gives you additional help, extra help. . . . it actually gives you more support for your body, thenecessary nutrients for your body because you don’t have enough from the food.” CA, Asian American,nonsupplement user, non-food stamp recipient

There is uncertainty about the role or purpose of a supplement .“It helps somehow [to] control the disease or something like that. . . . It’s for your memory, and you can go tosleep easier. It’s very good if that person is a woman and if she is pregnant . . . . [I]t is very good for her child . . .and for the elderly. Oh, I don’t know, I am not sure.” CA, Asian American, nonsupplement user, non-food stamprecipient


Perceptions of Health StatusThe participants were asked a series ofquestions about what constitutes goodhealth, their perceived health status, andany healthful or unhealthful activitiesthey engaged in. They were asked, “Doyou consider yourself to be healthy?”(table 1). In general, the womendescribed health status along acontinuum of well-being, with roughlyequal numbers expressing these threeviews: (1) they did not feel healthy,(2) they were somewhat healthy butcould make changes to improve theirhealth, and (3) they were healthy.

The primary difference between thenon-food stamp recipients and thecurrent or former food stamp recipientsis that the latter group reported morehealth problems, regardless of supple-ment use. Some women said they werenot healthy because of chronic diseasessuch as diabetes, hypertension, obesity,asthma, and arthritis. Some also indi-cated that they had epilepsy, anemia,gastrointestinal problems, mental healthconditions related to depression, andhistories of substance abuse anddomestic violence.

Reasons for Using NutrientSupplementsParticipants were asked whether theyconsumed anything besides food, forany reasons. Questioning was done toprobe for the range of possibilities ofsupplement use. They were then askedwhether they took any vitamin ormineral pills, and if they answered“yes,” they were asked what they took(either generic or brand names wereoffered), the dosage, and how oftenthey took the vitamin or mineralsupplement. They were also asked thereason(s) for taking supplements, themeans by which they obtained them,the estimated cost of the supplements,how they were paying for them (e.g.,out-of-pocket cash or insurancereimbursement), and any other

information about dietary supplements(e.g., herbs) that they and/or otherhousehold members were taking.The responses were categorized intoeight emergent themes that related to

1. Brands of supplements used byadults

2. Supplement use by children3. Reasons for use of single-nutrient

supplements4. Dosage of supplements5. Income constraints and patterns of

supplement use6. Acquisition of supplements7. Promotion of supplement use by

influential figures8. Media influence on supplement use

Multivitamins/multiminerals were themost common nutrient supplementtaken by the participants. Familymembers usually took the same brand.Some women and family memberstook supplements with added nutrientsbeyond a standard multivitaminformulation.

Children were most often given eithermultivitamins or nutrient supplementscontaining specific micronutrients suchas vitamin C and zinc. Parents who didnot take supplements themselves oftenensured that their children took a dailymultivitamin.

“My husband doesn’t [takevitamins], but my kids take ageneric multivitamin with extravitamin C; . . . off the top ofmy head, I don’t know [howmuch vitamin C], but theyeach take one of them. I justassumed that it would be betterfor them because they have somany different choices, andI just hear so much aboutvitamin C being so importantfor people. . . that’s why Igrabbed that one. . . . [I]t wasjust something I thought theyneeded, and of course, just

about everything I buy isgeneric because when you areon a low income like we are,you have to stretch your dollarsas far as you can. Sometimesit’s hard, but I just thought . . .they’ve all been healthy. . . .I’ve been pretty lucky.” AR,White, nonsupplement user,current food stamp recipient

Single nutrient supplements were takeneither in addition to or instead of amultivitamin and often for specificreasons. Some women said it wasimportant to purchase a multivitaminand mineral supplement if they thoughtit would provide a positive benefitsuch as improving their energy level,managing stress, building up theirbody reserves, preventing infections,or managing chronic diseases. Someparticipants were able to describewhy they were taking supplements asillustrated in the example below. Otherswere unable to describe clearly whatthe supplements were supposed to dofor them.

“I take a lot of herbs. I takeEchinacea, calcium, magnesiumwith zinc, and I take 1,000 mgof vitamin C every day. I [also]take some beta-carotene.”[Interviewer: Why are you takingthe calcium, magnesium, andzinc?] “Because it builds bones. . . . I was a polio victim when Iwas 3 months old and so I take itto build up [my bones] . . . andthen it’s good for . . . what doyou call it? Osteoporosis. Thevitamin C keeps colds outbecause, as a diabetic, you cancontract anything quickly. . . .[I]t stops the flu and colds andstuff. I took it through the winterand didn’t have any problem.AR, African American,supplement user, non-foodstamp recipient

2002 Vol. 14 No. 2 27

Table 1. Beliefs about personal health among food-stamp eligible women: “Do you consider yourself to be healthy?”

Perceived status Illustrative statements Respondents’ characteristics

“I am not healthy . . .” “Not me because I get so tired at the end of the day. NY, White, supplement user,See, I work a lot, but I get so tired at the end of the day. non-food stamp recipientI don’t know if this is normal. . . . It’s like when myhusband come[s] home at night and he’s trying even totalk to me, I can’t even open my eyes to talk to him.”

“Oh, no, because I’m overweight. I’ve been that way AR, White, nonsupplement user,pretty much all my life, and I think it’s not due to what food stamp recipientI eat. . . . it’s because of what I like to do. I don’t getenough exercise, that’s the biggest part. If I exercised,I could be the size I wanted to be, but there never seemsto be enough time in my day to take that time out for me.I eat stuff that’s not considered healthy. I don’t sit downand eat junk food like potato chips, but I don’t eat whatyou would call good-for-you foods like fruits and vegetables.We eat them, but I try to make it more a part of my kids’[diet] than I do mine. . . . I think about them, but I don’tstop to think about myself.”

“I am somewhat healthy . . . ” “Yea, pretty much. I might contradict myself here because AR, Latina, supplement user,I know that I’m overweight. . . . I don’t feel that it is food stamp recipientcausing me to be really unhealthy where I am dragging. . . .I could probably lose the weight and feel better.”

“No, because I don’t really eat right. I eat about one meal CA, White, nonsupplement user, a day sometimes, and then I will . . . snack throughout food stamp recipientthe day on chips and soda . . . you know, junk food. Idon’t eat right. . . . I consider myself to be somewhathealthy . . . healthier than . . . this is what I am trying to say.If I took vitamins, I don’t think that would changeanything.”

“I am healthy . . . ” “Yea. I consider myself to be healthy. I eat the right types AR, White, nonsupplement user,of foods, I hardly ever get sick with the flu or anything food stamp recipientlike that, and I exercise.”

“Yes . . . I’m a little overweight, but it’s okay. . . . I don’t NY, African American, supplementhave diabetes. . . . I haven’t developed any of those user, food stamp recipientdiseases. I don’t have heart disease yet. I’ve been tryingreal hard to keep it down. I’m trying to lose more weight.”


“. . . I thought I had ovariancancer but it was [endometriosis].I’m starting to feel better now . . .and I’m taking vitamins, whichI don’t like to do. . . . I don’tlike taking pills. I started 2months ago. My nails are gettingstronger, I’m feeling healthier,and my hair is growing faster.I take them every day. I love itbecause it is about $5 for a 30-day supply. It’s got the vitaminE, the magnesium and zinc, theherbal energy, and the rest ofthem . . . the value pack. . . . Ipay my own cash for them.”AR, White, supplement user,former food stamp recipient

Many participants were not alwaysattentive to the dosage or brand of thesupplements taken: They admittedtaking less than what was recommendedor not taking the supplements daily.Because these participants had limitedincomes, they wanted the supplementsto last longer.

“I take vitamin E for skin,vitamin C, and calcium. On the[vitamin E] bottle it says to takeone pill three times a day but Ionly take one a day because Ican’t afford to buy ’em for threetimes a day.” AR, AfricanAmerican, supplement user,current food stamp recipient

Most women living on limited incomeseither paid for supplements themselves,received them through Medicaid orMediCal when a prenatal multivitaminor iron was prescribed, or receivedthem from friends or relatives whowould share their supplements orpurchase supplements for the partici-pants when resources were low.

“The prenatal vitamins . . .when I ran out, I just didn’ttake them [any] more. . . . They

gave them to me free at theclinic . . . through MediCal. . . .The kids take vitamin B,vitamin C, and the little kidvitamins. . . . I pay for themout of my own pocket.” CA,African American,nonsupplement user, foodstamp recipient

Friends, relatives, and/or physiciansmost commonly recommendedsupplements. Pharmacists, dentists,and sales associates in health-foodstores were identified less frequently asauthorities encouraging supplement use.No participant identified a nutritionistor dietitian as a professionalrecommending supplement use.

The media was cited less frequentlythan were authoritative figures forinfluencing supplement use and wasreported to have both a positive andnegative influence on women’s useof supplements. In some cases, themedia messages influenced them to trysomething new. In other instances, themedia messages promoting supplementuse were disregarded, because thewoman questioned the benefit of theproducts.

“I think they are just trying toget you to buy the product . . .just like any commercial. Forsome people, it might be a goodthing. I don’t drink milk, somaybe I could take some typeof calcium supplement. But ifyou eat right and do everythingright, there is no need for that.They just want your money.”CA, White, nonsupplementuser, current food stamprecipient

One participant equated the side effectsof medications with the potential sideeffects of nutrient supplements andstated that she avoided them.

“Yea, we’ve seen [the TVadvertisements]. Well, theyshow all those side effects . . . ,and that scares me. . . . [S]ideeffects scare me to death. I tooksome antibiotics when I wassick, and I had some real badside effects. [Interviewer: Isthat different from a vitamin?]It’s just the side effects thatscare me.” AR, White,nonsupplement user, currentfood stamp recipient

Reasons Why Women Do NotTake Nutrient SupplementsSeveral different themes were identifiedto explain why women chose not totake nutrient supplements. About one-quarter of the participants believed itwas possible to get all the vitaminsand minerals one needs from food.

“I don’t take any vitaminsbecause I get all the vitaminsI need from the fruits andvegetables I eat.” AR, White,nonsupplement user, foodstamp recipient

Women and/or their children avoidedor discontinued supplement use forreasons such as cost and the need toprioritize expenses; side effects suchas nausea, dizziness, or constipation;potential or perceived side effects;and dislike of the taste by children.

“My kids will not take vitamins. . . . [T]hey don’t like the taste. . . . [The vitamins] taste nasty[or] have a funny taste. . . . [I]t’snot like regular foods that youcan prepare differently.” CA,Latina, nonsupplement user,current food stamp recipient

2002 Vol. 14 No. 2 29

The women cited several reasonsfor discontinued supplement use: amultivitamin could overstimulatethe appetite, the supplements hadpreviously not produced the anticipatedeffects, and for some women whowere already taking pills for medicalconditions, they did not want to takemore pills. A few did not think aboutpurchasing a multivitamin pill ornutrient supplement while groceryshopping.

The Use of Food Stamp Benefitsto Purchase NutrientSupplementsThe responses of participants weredivided into two groups concerning theuse of food stamps to purchase nutrientsupplements (table 2). The first groupconsisted of a minority of participantswho believed that food stamps shouldbe used only for food because (1) themonthly food stamp allowance was notadequate to meet a household’s foodneeds especially in large families,(2) recipients should eat vegetablesor fruits rather than take pills, and(3) a vitamin pill would not alleviatehunger or promote satiation as foodcould. The second group believedcertain circumstances deservedconsideration so that needy familiescould purchase nutrient supplements.

Several themes were identified tocharacterize the view of both foodstamp recipients and nonrecipientswho said it was a good idea to allowrecipients to purchase a multivitaminand mineral pill with their food stampbenefits because it might (1) assist themin getting what they need nutritionallywhile living on a low income; (2) helpparents save pocket money that couldbe used toward something else such asbuying children’s clothes or schoolsupplies; and/or (3) improve foodstamp recipients’ overall health.

In Arkansas, three food stamprecipients expressed that taking amultivitamin would be less expensivethan buying fresh fruit. Although theywould have preferred to purchase fruit,they believed that taking vitamin Cor a multivitamin would be the mostpractical and least expensive alternativefor low-income families.

Some said that changing the FSP policywas a good idea if recipients could notbuy healthful food. However, otherssaid it was a good idea because theybelieved that food stamp recipientsgenerally do not eat healthful foods.Other attitudes and beliefs sharedconcerning the benefits of supplementsincluded these:

• Food stamp recipients should takeone multivitamin instead of severalvitamin or mineral pills.

• It is feasible to use food stampbenefits for supplements ifrecipient makes wise budgetingdecisions.

• Supplement use would depend onthe person or family situation.

• Supplement use could set a goodexample for children and mightstimulate other healthful habitssuch as buying more healthfulfoods.

• It is easier to take a pill than to eathealthful food.

• Food stamp recipients need to beconvinced of the benefit of takinga multivitamin and mineral pillregularly.

A few food stamp recipients suggestedthat the government offer a specialcoupon to families each month thatcould be used to purchase a designatedsupplement—similar to providingspecific WIC commodities—but ifrecipients did not use the coupon, theywould lose the benefit.

Most women living on limitedincomes either paid forsupplements themselves,received them through Medicaidor MediCal . . . , or received themfrom friends or relatives . . . whenresources were low.


Table 2. Food stamp-eligible women’s perceptions regarding the use of food stamps to purchase nutrient supplements

Perceived status Illustrative statements Respondents’ characteristics

Group 1“Food stamps are for “I would rather buy food, because I get hungry and I need CA, White, nonsupplement user,food only because . . .” to eat . . . I like to eat. [A vitamin and mineral pill] won’t current food stamp recipient

do [anything] for me. . . . [T]he welfare office does not giveyou enough food stamps to have that luxury to also buyvitamins. I only get $230 for my two daughters, and theyare thinking that’s enough for food for the whole monthand it [isn’t]! You really don’t have that much moneycoming in to afford to buy that. You would rather haveyour kid eat food than a vitamin; . . . it would not be badif a person receiving food stamps also had more incomecoming in if that is what they want to do.”

“You buy vitamins automatically with food stamps, NY, African American, supplementbecause you’re buying your fruits and your vegetables user, current food stamp recipientand stuff like that . . . so it’s really the same thing.”

Group 2“Food stamps could be “It would help because then that money I spend on my AR, White, supplement user, currentused to purchase calcium, I could spend on something else. My kid food stamp recipientnutrient supplements always needs socks and underwear . . . he’s growing sobecause . . .” fast . . . .Yea, I would probably [take] vitamins and

[my son] would [take] vitamins, but vitamins and stufflike that are just outrageous. You just can’t afford it![If money wasn’t an issue], I would probably buy [vitamins]to make sure I was getting what I was supposed to and whatmy body really needed, so that my body wouldn’t breakdown, and I wouldn’t have so many health problems.”

“I think that would be great, because a lot of people out AR, White, nonsupplement user,there can’t shop, and it would be just as easy to take a pill current food stamp recipientto stay healthier. Many times I’ve been low on food stamps,cooked for the kids, and went hungry. . . . I could haveused that vitamin supplement.”

“I think a lot of your healthier foods are more expensive AR, White, nonsupplement user,. . . . [I]f you go to buy your fruits, other than bananas and current food stamp recipientapples and oranges, when you start buying for a family ofseven, you’re talking several bags of each. I think that to behealthier, you are going to have to spend more. . . . I thinkit’s too expensive to eat what they should eat.”

2002 Vol. 14 No. 2 31

Perceived Drawbacks to UsingFood Stamps to PurchaseNutrient SupplementsSeveral participants shared somepossible drawbacks to allowing foodstamp recipients to use their benefits topurchase supplements. They believedrecipients might purchase supplementsbut not take them, might not give theirchildren adequate food if householdresources were spent on a supplement,might not be able to absorb thenutrients from a pill or may be allergicto the supplement, or might abuse theFSP by selling food stamp benefits ornutrient supplements for cash. Theyalso thought that allowing recipients topurchase supplements might reinforcethe perception that food is not needed ifvitamins are substituted and that takingtoo many or high doses of supplementsmay be harmful.

Decisionmaking About theUse of Food StampsParticipants were asked their opinionregarding who should decide how foodstamps are used—either the governmentor recipients. Three themes emerged:they believed food stamp recipientsshould decide, the government shoulddecide, or the government and peopleshould work together to decide. Morethan half of the food stamp recipientsindicated that the people rather than thegovernment should decide how foodstamps are used. Many acknowledged,however, that the government’s positionwould be more heavily weightedbecause it provides the benefits.

Discussion

The FSP-supplement proposal, far frombeing a simple policy change, bringstwo relatively new concerns to theforeground with respect to the goalsof the FSP: (1) Should the goals of the

FSP be broadened to include healthpromotion beyond that associated withachieving equity in food intake? Arelated question is whether the supple-ment proposal is an appropriate strategyfor doing so. (2) Given the high preva-lence of supplement use in the generalpopulation, should food stamp recip-ients have the same level of choice asthe general population regarding howthey obtain their nutrients (i.e., viafoods or supplements)? A relatedquestion is whether the current FSPpolicy constrains such choice. Thislatter question reveals a concern forconsumer autonomy as distinct fromequity or health promotion. Autonomyhas not been one of the stated goalsof the FSP; for example, currentregulations do not permit the use offood stamps to purchase preparedfood away from home.

While equity, health promotion, andautonomy all are implicated in thisissue, much of the debate has empha-sized only one or another of thesegoals and has not examined the actualstrength of the trade-offs among them.The findings from the present studyare synthesized below, in order toshed light on these policy questions.

As shown in figure 1, women in thisstudy appear to hold an overallphilosophy regarding nutrient supple-ments that is shaped by their beliefsconcerning the nutritional adequacyof food, the inadequacy of actualbehaviors, perceived benefits andexperiences, the concept of supple-ments itself, and their current supple-ment practices. This philosophyappears to be malleable and/ornegotiable depending upon suchfactors as degree of self-reflection, theclarification of existing information oraddition of new information (especiallyfrom influential interpersonal sources),or changes in the participants’ healthstatus or income. In a few cases in thisstudy where participants believed their

diet was adequate to provide themwith all the nutrients they needed,their general philosophy appearedless malleable.

The most plausible prediction to bederived from these findings is thatnutrient supplement use will increase inthis population as long as the dominantnarrative in their personal informationnetworks and other influential sourcesis positive toward using nutrientsupplements. Conversely, informationfrom national authorities—as reportedin the media or through programs suchas FSNEP or EFNEP—concerningadverse events, lack of efficacy, or falseadvertising claims is unlikely to reachthis population efficiently, although itmay do so after an indeterminate lagtime. Supplement use is predicted toincrease under both the existing FSPpolicy and revisions in the policy,although it is likely to be more rapidand extensive under a changed FSPpolicy.

Implications for Policy GoalsIn attempting to relate these predictionsto equity in food access, healthpromotion, and personal autonomy,we find it necessary to consider thepotential effect of increased use ofnutrient supplements on total householdexpenditures, on the quality of foodintake itself, and on other health-relatedbehaviors. Moreover, it is necessary toexamine more carefully the meaning ofautonomy in light of the informationasymmetries noted in the precedingparagraph.

Effect on Food Access. With regardto household expenditures, the expecta-tion derived from household economicsis that expenditures for supplementswould reduce the income availablefor all other expenditures by an equalamount. That is, it would be subtractedfrom the total household budget notexclusively from the household food


Figure 1. Conceptual framework for food, health, and nutrient supplements among low-income, food stamp-eligible women

budget because only a fraction of FSPhouseholds (25 to 40 percent) currentlypurchase their entire food supply withfood stamps (USDA, 1999). Most(60 to 75 percent) augment their foodstamps with other income sources,indicating a substitutability betweenfood and non-food expenditures, andthey must do so to an even greaterextent when they purchase supplements.This is true regardless of whether foodstamps or cash is used to purchase thesupplements.

A high-end figure of $5/month (17cents/day for a mother and two childrenas used by USDA [1999] and theaverage food stamp household of 2.4persons in 1999 [with $338/month netincome plus $162 in food stamps]), forexample, would represent a 1-percentdecrease in income available for allother expenditures. If all of this weresubtracted from non-food expenditures,there may be no effect on food expendi-tures. If all of it were subtracted fromfood expenditures (which average $224/

month for FSP households), the effectcould be a 2.4-percent reduction ($5)in food expenditures.

Using a different set of assumptions andmethods, USDA estimated a low-endeffect of 26 cents/month and high-endeffect of 94 cents/month (0.4 percent)reduction. One of the major reasons forthis difference (compared with thepresent estimates) is that the USDAmethod averages the effect across allFSP households; whereas, the presentmethod emphasizes the potential effectson smaller subgroups (USDA, 1999).Notwithstanding these large differencesamong estimates in percentage terms,effects of this size do not appear toconstitute a significant threat to foodaccess, especially since householdswould retain the option of foregoingsupplements in favor of purchasingfood.

However, the policy change couldhave more serious implications for foodaccess if it were to lead eventually to

changes in the way benefit levels arecalculated. Specifically, if the baseassumption for future Thrifty Food Plancalculations is that nutrient supplementscan be used to meet some or all of aFSP recipient’s vitamin and mineralrequirements, especially for those thatare relatively expensive from foodalone such as folate, this could lead tosignificant reductions in benefit levelsand, subsequently, food access. This isnot a minor policy consideration.

Effect on Health Promotion. In theory,a change in the FSP policy could affectrecipients’ health in several ways. Abenefit is that it could compensate foror enhance the vitamin and mineralintake of recipients who have un-healthful diets. A drawback is that itcould compromise the quality of foodchoices and intake because of thebelief that nutrient supplements are aneffective substitute for food. Reports bythe USDA and Life Sciences ResearchOffice address the former possibilityin considerable detail (Life Sciences

Current Supplement Practices

Adequacy of

Food Itself

(majority view)

Inadequacy of Actual Behavior (majority view)

fast-paced society, income, food preferences, disease, perceived health, other health-promoting behaviors, knowledge, processed foods

Concept of “Supplement”

replacement/substitute (majority view), addition/complement, unclear

Perceived Benefits and Personal Experiences (variable)

energy, immune function, illness prevention, side effects, hard to swallow, poor taste, children (unspecified reasons)

Self-reflection, existing information, new information, influential sources, change in personal circumstances, sociocultural influences

Philosophy Regarding

Supplements

(often malleable)

Beliefs

Influences

2002 Vol. 14 No. 2 33

Research Office, 1998; USDA, 1999),and the present study does not addfurther insight into the findings.However, the present study doessuggest the danger that greater supple-ment use in this population may leadto compromises in the quality of foodchoices and intake—especially if thereis an implied government endorsementof supplement use as a result of achange in food stamp policy.

Implications for Autonomy. Themajority of study participants expressedthe view that food stamp recipientsshould not only make their own choicesregarding supplement use (citingnormative as well as pragmatic reasons)but also that the government shouldcreate the circumstances that wouldsupport recipients’ choices. While thesereasons have strong support from theperspective of ethics and welfareeconomics, respectively, this studyalso reveals some countervailingconsiderations that demand equalattention.

First is the documentation of a widerange of misconceptions concerning thepurpose and role of a supplement, thenature of its benefits, and the degree towhich the health benefits of food andsupplements are substitutable. Even ifthere is no change in the current policyregarding supplements, these mis-conceptions deserve attention in currentnutrition education programs for low-income populations. Second is thedocumentation that this population islikely to face significant informationasymmetries because of the nature oftheir influential information sources,and will be unable to discover hiddenquality defects on their own suchas lack of efficacy and adverseconsequences.

Both of these situations suggest amiddle ground in which food stamprecipients could exercise autonomy indecisionmaking, but government should

take effective steps to correct potentialinformation failures. Such steps wouldneed to include the following: design-ing a clear and effective educationinitiative regarding the use of nutrientsupplements; regulating labeling,advertising, and other forms of pro-motion based on the messages that aretargeted for and understood by thispopulation; and evaluating the extentto which an implicit governmentendorsement of multivitamin andmineral supplement use (and itsassociated promotion by the supple-ment industry and retailers) is general-ized by members of this population toinclude higher doses and/or other formsof dietary supplements. (This latterpossibility was not investigated directlyin this study.) Finally, a policy changeregarding supplements would require avariety of administrative changes todefine eligible items; inform manu-facturers, retailers, and consumers ofthese rules; and monitor and enforcecompliance with these rules.

Strategies for Improving Foodand Nutrient IntakesIf promoting the health of low-incomeAmericans beyond that required forachieving equity is deemed a worthypolicy goal, attention should then focuson the most effective and appropriatestrategies to do so. Previous discussionshave explicitly noted the logical fallacyof assuming that the most effectiveand appropriate strategy necessarilyinvolves supplements or even the FSPitself (Life Sciences Research Office,1998). Instead, this earlier analysisconsidered supplements, fortification,a variety of other incentives, andpromotional or enabling strategies toimprove food and nutrient intake topromote good health among food stamprecipients and low-income people ingeneral. But even this is only a partiallist of the potential strategies forpursuing one of the core public healthgoals as outlined in Healthy People

2010 (U.S. Department of Health andHuman Services [DHHS], 2000).

Investigating the larger food environ-ment of food stamp recipients wouldbe appropriate for the purpose of pro-moting health and other food-systemgoals such as improving access to andthe quality of supermarkets, supportingthe capacity of institutions that servelow-income groups to purchase freshproduce from local farmers, andexpanding the ability to use electronicbenefit transfer cards at farmers’markets or for community-supportedagricultural schemes. A systematicexamination of potential strategies(and an effort to reconcile health, foodsecurity and food-system goals pursuedby other government programs) doesnot appear to have been undertaken.

Conclusions

One version of the FSP-supplementpolicy dialogue maintains that a changein FSP policy would give programparticipants the same freedom to usenutrient supplements that otherAmericans have and improve theirhealth and nutritional status at a lowercost than is possible through carefulfood selection. A decision to changethe FSP policy based on this narrativewould greatly overestimate the benefitsassociated with a multivitamin andmineral supplement in a populationwhere nutrient deficiencies are rare.It would overlook the potential fornegative consequences (i.e., decreasingthe intake of nutrient-dense foods),and it would neglect the imperfectionsand asymmetries in the informationavailable to food stamp recipients andthe cost of government actions requiredto correct this class of market failureseffectively.

A decision to change FSP policy furthersuggests that the net effect of pro-ceeding with the policy change in the


absence of effective actions to correctfor these market failures would be toshift some additional costs or risks ontoan already disadvantaged populationfor the sake of little additional benefit,thereby raising serious ethical concerns.

This study reveals the need to conducta more systematic examination of thepotential strategies for improving thenutritional health of food stamp-eligiblehouseholds and the importance of usinga more complete list of criteria whenattempting to identify the most effectiveand appropriate goals and strategies.

Acknowledgments

This research was funded through theFood and Nutrition Research smallgrants program sponsored by theUSDA Economic Research Serviceand administered by the University ofCalifornia at Davis. We are grateful tothe Cooperative Extension staff in NewYork City; Fort Smith, Arkansas; andSan Francisco, San Jose, and Oakland,California, for their generous assistancein conducting this study.

References

American Dietetic Association. (2000). Nutrition and you: Trends 2000. Journalof the American Dietetic Association, 100, 626-627.

Balluz, L.S., Kieszak, S.M., Philen, R.M., & Mulinare, J. (2000). Vitamin andmineral supplement use in the United States: Results from the Third NationalHealth and Nutrition Examination Survey. Archives of Family Medicine, 9, 258-262.

Bender, M.M., Levy, A.S., Schucker, R.E., & Yetley, E.A. (1992). Trends inprevalence and magnitude of vitamin and mineral supplement usage andcorrelation with health status. Journal of the American Dietetic Association, 92,1096-1101.

Dickinson, A. (1998). Optimal Nutrition for Good Health: The Benefits ofNutritional Supplements. Washington, DC: Council for Responsible Nutrition.

Hunt, J. (1996). Position paper of the American Dietetic Association: Vitamin andmineral supplementation. Journal of the American Dietetic Association, 96, 73-77.

Johnson, R.K., & Kennedy, E. (2000). The 2000 Dietary Guidelines forAmericans: What are the changes and why were they made? Journal of theAmerican Dietetic Association, 10, 769-774.

Kraak, V., Pelletier, D., & Dollahite, J. (2000). How Do Food Stamp EligibleWomen Negotiate Food and Nutrient Supplement Choices? Qualitative FindingsFrom an Ethnically and Regionally Diverse Sample of Low-Income NutrientSupplement Users and Nonusers. Division of Nutritional Sciences, CornellUniversity, Ithaca, New York (unpublished).

Life Sciences Research Office. (1998). Analysis and Review of Available Data andExpert Opinion on the Potential Value of Vitamin and Mineral Supplements toMeet Nutrient Gaps Among Low-Income Individuals. Draft report prepared for theUSDA Economic Research Service, Bethesda, MD (unpublished).

Miles, M.B., & Hubberman, A.M. (1994). Qualitative Data Analysis. ThousandOaks, CA: Sage Publications.

Nayga, R.M., & Reed, D.B. (1999). Factors associated with the intake of dietarysupplements. Family Economics and Nutrition Review, 12, 43-48.

Neuhouser, M.L., Patterson, R., & Levy, L. (1999). Motivations for using vitaminand mineral supplements. Journal of the American Dietetic Association, 99, 851-854.

Newman, V., Rock, C.L., Faerber, S., Flatt, S.W., Wright, F.A., & Pierce, J.P.(1998). Dietary supplement use by women at risk for breast cancer recurrence.Journal of the American Dietetic Association, 98, 285-292.

2002 Vol. 14 No. 2 35

Office of Dietary Supplements, National Institutes of Health. (1999). RetrievedSeptember 12, 2000, from http://ods.od.nih.gov.

Pelletier, D.L. (2001). Research and policy directions. In R. Semba & M. Bloehm(Eds.), Nutrition and Health in Developing Countries. Totawa, NJ: Humana Press.

Pelletier, D., & Kendall, A. (1997). Supplement use may not be associated withbetter food intake in all population groups. Family Economics and NutritionReview, 10, 32-44.

Porter, D.V. (1995). Purchasing dietary supplements with food. Nutrition Today,30, 214-217.

Skolnick, A.A. (1995). Experts debate food stamp revision. Journal of theAmerican Medical Association, 274, 781-783.

Slesinski, M.J., Subar, A.F., & Kahle, L.L. (1995). Trends in use of vitamin andmineral supplements in the United States: The 1987 and 1992 National HealthInterview Surveys. Journal of the American Dietetic Association, 95, 921-923.

Subar, A.F., & Block, G. (1990). Use of vitamin and mineral supplements:Demographics and amount of nutrients consumed. American Journal ofEpidemiology, 132, 1091-1101.

Thomas, P. (2000). Food stamps for dietary supplements? The DietarySupplement, Issue 1(9). Retrieved September 12, 2000, from http://www.thedietarysupplement.com.

U.S. Department of Agriculture. (1999). The Use of Food Stamps to PurchaseVitamin and Mineral Supplements. Retrieved September 12, 2000, from http://www.fns.usda.gov/oane/menu/Published/fsp/FILES/Program%20Design/vitamin.pdf.

U.S. Department of Health and Human Services. (2000). Healthy People 2010.Washington, DC.

2002 Vol. 14 No. 2 36

Factors Affecting Meat PreferencesAmong American Consumers

This study analyzed socioeconomic and nutritional factors affecting consumers’preferences for meatless meals and, specifically, meals with less red meat.Consumers’ preferences were influenced by geographical location, racial andethnic background, family composition, and household income. Although femaleswere not statistically different from males in their preference for red meat, theygenerally preferred more meatless meals. Similarly, older respondents preferredmore meatless meals and less red meat than did younger respondents. Meat wasless preferred, as well, among American households in the highest income group.Advanced educational level of the respondents positively influenced the prefer-ence for meatless meals and meals with less red meat, and nutritional concernsamong consumers also had a significant influence. Notably, compared withconcerns for vitamins and minerals, concerns for cholesterol, fat, and sugar had astronger effect on the preference for less red meat. Data such as those presentedhere can provide useful descriptions of the distribution of health-influencingbehavior in our population.

mericans are consuming lessred meat such as beef and morenon-red meats such as poultry.

In 1999 the per capita consumptionof red meat was 117.7 pounds, an 11-percent drop since 1970 (U.S. Depart-ment of Agriculture [USDA], 2001a;USDA, 2001b). Poultry consumption,however, increased 102 percent duringthe same period (up to 68.3 pounds perperson). In general, trends in consump-tion of animal products during the last30 years involve more use of poultry,fish, lowfat milk, yogurt, and cheeseand less use of red meat, whole milk,eggs, butter, and lard (USDA, 2001a).Although the decrease in the consump-tion of red meat has been compensatedby an increase in the consumption ofpoultry to a certain degree, overallconsumption of red meat has declined.Price, income, taste, and preferencesare the key variables affecting the levelsof meat consumption (Putnam &Gerrior, 1997).

Differences in retail price between beefand poultry may explain some of thedecline in the consumption of red meat.For example, per pound prices of retailbeef and poultry have remained at anaverage of $3.70 and $1.50, respec-tively (Bureau of Labor Statistics,2001). Changes in income have rela-tively little or even a negative effect onthe demand for red meat. For example,a report on food spending in Americanhouseholds in 1997-98 revealed thathigher income groups, compared withthe middle-income groups, decreasedtheir total spending on beef (Blisard,2001).

Besides relative prices and income,many other factors played key rolesin changing the demand for red meat.According to a report by the U.S.Department of Agriculture’s (USDA)Economic Research Service (ERS)(Putnam & Allshouse, 2001), consumerconcern about cholesterol and saturatedfat, inconsistent quality, and lack of

Arbindra P. Rimal, PhDSouthwest Missouri State University

A

2002 Vol. 14 No. 2 37

Table 1. Change in consumption of meatless meals and red meats

“You are eating more “You are eating less meatless meals than beef, pork, or lamb than

Response you used to” you used to”

PercentStrongly disagree 34.97 38.52Somewhat disagree 19.57 22.30Neither agree nor disagree 7.84 9.23Somewhat agree 15.36 11.36Strongly agree 22.26 18.59

n = 2,880.

convenience in preparation are associ-ated with the negative trend in beefdemand. Increasingly, associationsbetween consumption of red meat andthe onset of chronic disease have beenreported. Examples include linksbetween (1) metastatic prostate cancerand intakes of red meat and dairyproducts (Michaud et al., 2001) and(2) colon and other types of cancerand high consumption of red meatcombined with low intakes of dietaryfiber, fruits, and vegetables (Law,2000).

Additional research has shown thatpeople reduce their risk for coloncancer when they substituted lowfatdairy products for high-fat versions,margarine for butter, poultry for redmeat, and whole grains for refinedgrains (Slattery, Boucher, Caan, Potter,& Ma, 1998). These findings exemplifythe enormous body of literature linkingpatterns of overall dietary intake withincreased risk for cancer and otherchronic diseases.

Specific dietary patterns that beginduring childhood—such as the con-sumption of high-fat dairy productsand red meats—are likely to increaseage-specific rates of cancer and otherdiseases in adult life; however, the riskmay be reversed with later dietarychange. For example, a reverse inchildhood dietary patterns is demon-strated by more younger Americansbecoming vegetarians. About 15percent of the 15 million U.S. collegestudents eat vegetarian meals during atypical day (Walker, 1995). In a similarstudy of 158 British undergraduatestudents, more females than males hadavoided red meat. The main reasongiven was related to concerns for healthand sensory factors (Santos & Booth,1996).

Our study analyzed socioeconomic andnutritional factors affecting consumers’preferences for meatless meals and,

specifically, meals with less red meat.Nutritional concerns related to meatwere examined. A regression analysiswas performed to identify statisticallysignificant socioeconomic and demo-graphic characteristics, as well asnutritional concerns influencingconsumer preference.

Methods

Data and SampleThe data set used in this study was partof a nationwide telephone survey of2,880 U.S. households conducted bythe Gallup Organization in 1997 for theNational Peanut Association (NationalPeanut Council, 1997). The surveyused a probability sampling methodand included adults only. Gallup used amultiple-call-back method to eliminatebias in favor of those easy to reach bytelephone. A 95-percent confidenceinterval revealed a maximum expectederror range from the sample at ±3.1percent.

Survey questionnaires includedconsumers’ stated changes in meatconsumption (if any), as well asnutritional concerns while selectingfoods. Four market regions (West,Midwest, Northeast, and South) wereidentified, and these divisions wereconsistent with the designations of

geographical regions by the U.S.Bureau of the Census. Demographicsample means compared well withpopulation averages. For example,85 percent of the U.S. population isWhite; 51 percent is female. Thesample population was 84 percentWhite and 57 percent female. Theregional distributions of the sampleand the U.S. population were nearlyidentical.

VariablesThe dependent variables were devel-oped from consumers’ stated changesin meat consumption based on theirresponses to a 5-point scale: stronglydisagree, somewhat disagree, neitheragree nor disagree, somewhat agree, orstrongly agree. The change in overallmeat consumption was measured withthe statement: “You are eating moremeatless meals than you used to.” Thechange in the consumption of red meatwas measured with the statement: “Youare eating less beef, pork, or lamb thanyou used to.” Although most of therespondents either strongly disagreedor somewhat disagreed that they wereeating more meatless meals or less redmeat, more than 38 and 30 percentsomewhat agreed or strongly agreedthat they were eating more meatlessmeals and less red meat, respectively(table 1). These percentages representan important change in the preference


Table 2. Socioeconomic status andnutrition concerns when selecting fooditems

Variables

PercentGeographic regionNortheast 19.7South 33.7Midwest1 25.5West 21.1

RaceWhite 83.6African American 7.4Other1 9.0

GenderMen1 43.1Women 56.9

Marital statusMarried 55.5Unmarried1 44.5

Household typeHousehold with children 37.5Household without children1 62.5

EducationNo college1 37.9Some college or higher 62.1

MeanAge 45.0

Household income $39,900

Nutrition concerns2

Cholesterol level in food3 5.6Amount of fat3 6.7Vitamins and minerals3 5.0Sugar3 5.1

1Reference group in the regression model;other consists of Asian Americans, Hispanics,and Latin Americans.2Scores ranged from 1 to 10, where 10 = veryconcerned.3Coefficient of variation: cholesterol, 60.23;fat, 47.59; vitamins and minerals, 60.58;and sugar, 61.33.n = 2,880.

for meat, particularly red meat, consid-ering the large percentage of respon-dents who were eating meat and dairyproducts. Gallup poll data showed that90 percent of the respondents normallyconsumed red meat, 93 percentconsumed fish and poultry, and 94percent consumed dairy products(National Peanut Council, 1997).

Geographic region, household income,race, gender, marital status, householdtype, education, and age of the respon-dents were the demographic variablesused in the regression models (table 2).About one-third (33 percent) of therespondents lived in the South, four offive (84 percent) were White, and morethan half (51 percent) were women.Over half were married (56 percent)and resided in households with children(53 percent). These respondents, onaverage, had attended at least somecollege, were 45 years old, and had agross annual household income justunder $40,000.

Respondents were asked the followingnutrition-related question: “When youchoose the foods you eat, please tellme how frequently you consider thefollowing issues, using a 10-point scale,where 10 means you consider nearlyall the time (NAT), and 1 means youalmost never (AN) consider it.” Thenutrition issues included cholesterol,fat, vitamins and minerals, and sugar.As expected, mean responses weregenerally neutral: On average, house-holds tended to consider both desirable(vitamins and minerals) and undesirable(cholesterol, fat, and sugars) nutritionfactors “sometimes” when makingfood-purchase decisions. However,the responses varied considerably.

Empirical ModelAn ordered probit regression modelwas used because the preferencevariable was measured with a scalethat allowed the outcomes to be ranked.

In our study, women respondentswere not statistically differentfrom their male counterpartsabout their preference for redmeat, although they preferredmore meatless meals.

2002 Vol. 14 No. 2 39

Table 3. Consumer preference for meatless meals: Results of ordered probit models

Prefer more meatless meals Prefer less red meat Standard StandardVariables Coefficient error Coefficient error

Constant -0.2840* 0.1749 -0.4243* 0.1420Northeast 0.1350* 0.0824 0.2269* 0.0683South -0.0363 0.0727 0.1092* 0.0588West 0.1396* 0.0820 0.3007* 0.0674White -0.0198 0.1175 0.0031 0.0880African American -0.0221 0.1515 0.2071* 0.1197Women 0.3374* 0.0611 0.2786 0.0481Married -0.1557* 0.0643 -0.0472* 0.0520Households with children -0.2442* 0.0677 -0.1612* 0.0524Some college or higher 0.1762* 0.0626 0.1528* 0.0506Household income 0.0044* 0.0016 0.0027* 0.0013Age 0.0070* 0.0021 0.0093* 0.0018Cholesterol level in food 0.0315* 0.0106 0.0312* 0.0088Amount of fat 0.0244* 0.0111 0.0240* 0.0093Vitamins and minerals 0.0227* 0.0107 0.0058 0.0085Sugar 0.0322* 0.0106 0.0231* 0.0087F (Threshold parameter 1) 0.5566* 0.0393 0.3680* 0.0239F (Threshold parameter 2) 0.7907* 0.0501 0.6102* 0.0304F (Threshold parameter 3) 1.4312* 0.0824 1.2136* 0.0443Log likelihood function value -3145.30 -3075.55Log likelihood function value (Restricted; β=0) -3263.37 -3213.44χ2 236.15* 275.79*Madalla’s Pseudo R2 0.10 0.12

*Significant at p < 0.10.

The empirical model is defined as

Y*t=$’Xt + ,t

where Y* t is an unobserved preferencefor meat; $ is the vector of unknownparameters; Xt is a vector of fournutritional concern variables andsociodemographic variables hypothe-sized to affect the overall preferencefor meat and red meat; and ,t is theindependently and identically normallydistributed error term. While Y* t isunobserved, respondents actually reportpreference by selecting one of the fivecategories (Yt) representing consumers’like or dislike of meat in general andred meat in particular. Values for Ytare 1 through 5, where 1 representsstrongly disagree and 5 representsstrongly agree to the statements: “Youare eating more meatless meals thanyou used to,” and “You are eating lessbeef, pork, or lamb than you used to.”The unknown parameter vector in theempirical model, $, was estimated byusing LIMDEP software (Greene,1995).

The results of ordered probit modelswere interpreted by using the partialchange or marginal effects on theprobability of ordinal outcome. Indoing so, the independent variables—other than the one being examined—were held constant at their mean values.

Results

Results from the ordered probitregression models for consumerpreference for meatless meals andless red meat are reported in table 3.In addition, regression models wereused to determine the marginal effectsof the independent variables onconsumers’ preferences for meatlessmeals and less red meat (tables 4 and5). The chi-square statistics for bothmodels rejected (at the 0.01 level) thenull hypothesis that all parameters were

jointly zero. Maddala’s pseudo R2, usedto evaluate the fitness of the models(Long, 1997), was 0.10 for the prefer-ence for meatless meals and 0.12 forless red meat. For cross-sectional datawith categorical dependent variables,the pseudo R2 is often small (Gujarati,1995). Based on diagnostic tests(Belsley, Kuh, & Welsch, 1980), nocollinearity problems were detected inthe analyses.

Consumers’ preferences for meatlessmeals and red meat were influenced bysocioeconomic variables and nutritionconcerns. For example, householdsliving in the Northeastern and WesternUnited States preferred more meatlessmeals and less red meat, compared withthose living in the Midwest (table 3).

Differences between Midwestern andSouthern States regarding the prefer-ence for meatless meals were notsignificant. The results are consistentwith the livestock-related economiesof the country. Large numbers ofhouseholds in the Midwest depend onthe livestock industry, particularly beef;this may influence preferences towardmeat and red meat in general. AfricanAmericans were less likely than otherracial groups (Asian Americans,Hispanics, and Latin Americans) toprefer red meat. A difference in attitudetoward meat and red meat consumptionassociated with differences in ethnicbackground was reported previouslyin a comparative study that tested forethnic differences in consumption ofdietary fat in a community-based


Table 4. Marginal effects of the independent variables on consumer preferencefor meatless meals

Disagree AgreeVariables Strongly Somewhat Neutral Somewhat Strongly

Constant 0.0695 0.0217 0.0044 -0.0068 -0.0888Northeast -0.0330 -0.0103 -0.0021 0.0033 0.0421South 0.0089 0.0028 0.0006 -0.0009 -0.0114West -0.0342 -0.0107 -0.0022 0.0034 0.0437White 0.0048 0.0015 0.0003 -0.0005 -0.0061African American 0.0054 0.0017 0.0003 -0.0005 -0.0069Female -0.0826 -0.0258 -0.0052 0.0081 0.1055Married 0.0381 0.0119 0.0024 -0.0038 -0.0486Households with children 0.0598 0.0186 0.0038 -0.0059 -0.0763Some college or higher -0.0431 -0.0134 -0.0027 0.0042 0.0550Household income -0.0011 -0.0003 -0.0001 0.0001 0.0014Age -0.0017 -0.0005 -0.0001 0.0002 0.0021Cholesterol level in food -0.0077 -0.0024 -0.0005 0.0008 0.0098Amount of fat -0.0060 -0.0019 -0.0004 0.0006 0.0077Vitamins and minerals -0.0056 -0.0017 -0.0004 0.0005 0.0072Sugar -0.0079 -0.0025 -0.0005 0.0008 0.0101

sample of Hispanic and White adultswith low educational attainment(Winkleby, Albright, Howard-Pitney,Lin, & Fortmann, 1994). The studyreported high consumption of dietaryfat among Whites with low educationalattainment and increased consumptionof fat among Hispanics at higher levelsof acculturation.

Family composition and children in thehouseholds significantly affected thehouseholds’ preferences for meatlessmeals and red meat. Households withchildren were likely to be concernedabout nutritional balance in the diet.For example, zinc deficiency is knownto occur in children’s diets that are lowin sources of readily bioavailable zincsuch as red meat and high in unrefinedcereals that are rich in phytates anddietary fiber (Sandstead, 1991).

In the study reported here, householdswith children and married-couplehouseholds preferred more meals with

meat and red meat than did the house-holds without children and unmarried-couple households, respectively (tables3 and 4). The probability that therespondents strongly agreed that theywere eating meatless meals was lowerby 5 percent among married respon-dents than that of unmarried respon-dents (table 4). Similarly, the differencein the probability of respondents inhouseholds with children, comparedwith their counterparts, stronglyagreeing that they were eating meatlessmeals was as high as 8 percent.

Similar results were reported in astudy among Australian married-couplehouseholds with children, where groupswith lower mortality rates tended tospend more money on fruits, veg-etables, cereal products, and fish,compared with groups with higherrates of mortality (Powles, Hage, &Cosgrove, 1990). In addition, house-holds with these lower mortality ratesspent substantially less on alcohol andsubstantially more on red meat.

Education level of therespondents positivelyinfluenced the preferences formeatless meals and red meat:Respondents with either collegeor an advanced level of educationpreferred to have more meatlessmeals and less red meat than didthose with less than a collegeeducation.

2002 Vol. 14 No. 2 41

Table 5. Marginal effects of the independent variables on consumer preference forless red meat

Disagree AgreeVariables Strongly Somewhat Neutral Somewhat Strongly

Constant 0.1140 0.0386 0.0177 -0.0016 -0.1687Northeast -0.0007 -0.0002 -0.0001 0.0000 0.0010South 0.0433 0.0147 0.0067 -0.0006 -0.0641West -0.0008 -0.0003 -0.0001 0.0000 0.0012White -0.0556 -0.0188 -0.0087 0.0008 0.0823African American -0.0411 -0.0139 -0.0064 0.0006 0.0608Female -0.0025 -0.0008 -0.0004 0.0000 0.0037Married -0.0748 -0.0253 -0.0116 0.0010 0.1107Households with children 0.0127 0.0043 0.0020 -0.0002 -0.0188Some college or higher -0.0610 -0.0206 -0.0095 0.0008 0.0903Household income -0.0808 -0.0273 -0.0126 0.0011 0.1196Age -0.0293 -0.0099 -0.0046 0.0004 0.0434Cholesterol level in food -0.0084 -0.0028 -0.0013 0.0001 0.0124Amount of fat -0.0064 -0.0022 -0.0010 0.0001 0.0095Vitamins and minerals -0.0016 -0.0005 -0.0002 0.0000 0.0023Sugar -0.0062 -0.0021 -0.0010 0.0001 0.0092

Household income positively influ-enced consumers’ preferences towardmore meatless meals and less red meat.Increases in household income hadpositive marginal effects on theprobabilities for other categories suchas “somewhat agree” and “stronglyagree” for more meatless meals and lessred meat (tables 4 and 5). That is, each$10,000 increase in annual householdincome increased the probability thatrespondents’ “strongly agreed” theywere eating more meatless meals andless red meat by 1.4 and 12 percent.The marginal effect is more impressivefor less red meat than it is for meatlessmeals.

This result agrees with findings ofthe 1997-98 USDA report on FoodSpending in American Households(Blisard, 2001), which showed thatduring the 2-year period of 1997-98,Americans in the highest income groupspent only $91.22 per person on beef,whereas the middle-income group spentonly $94.53. During the same period,the highest income group spent $193.73per person for all meat items. By com-parison, the middle-income group spent$196 for all meat items.

Nayga (1996) reported that incomehad significant and positive effects onconsumers’ use of information regard-ing undesirable nutrition factors such asfat, calories, and cholesterol. In anotherstudy, researchers showed that amongBulgarian households, concern aboutfat content in food items was positivelyrelated to their income (Moon et al.,1998).

Many studies have suggested a greaterhealth concern among women, com-pared with men (Frazao & Cleveland,1994; Nayga & Capps, 1994). In ourstudy, women respondents were notstatistically different from their malecounterparts about their preference forred meat, although they preferred moremeatless meals (tables 3 and 4). In

related studies, Lin (1995) noted thatfemales were more likely to believefood safety was very important in foodshopping; and Guthrie, Fox, Cleveland,& Welsh (1995) reported that femaleswere more likely to use nutritionallabels in making food selections.

Education level of the respondentspositively influenced the preferencesfor meatless meals and red meat:Respondents with either college or anadvanced level of education preferredto have more meatless meals and lessred meat than did those with less than acollege education. The marginal effectsof education on the dependent variables(tables 4 and 5) show that respondentswith some college or higher educationwere 5 and 9 percent more likely to“strongly agree” that they were eatingmeatless meals and less red meat,respectively, than were those withoutsome college or higher education.Presumably, respondents with a highereducation read nutrition informationand connected diet-disease relation-

ships (Nayga & Capps, 1999; Putler& Frazao, 1994). Thus, public healthand nutrition education can raise thenutritional awareness of consumers bytargeting those segments of the popula-tion with less education.

Our findings suggest that older respon-dents, compared with younger respon-dents, preferred more meatless mealsand less red meat. Positive relationshipsbetween age and general healthconcerns were reported in previousstudies. These include concerns forfood safety (Michaud et al., 2001)and using food labels for nutritionalinformation (Nayga, 1996).

Nutrition concerns that affect selectionof food items had a positive influenceon consumers’ preferences for meatlessmeals and less red meat. In this study,as the level of concern for nutritionincreased, consumer preferences formeat in general and red meat inparticular were likely to decrease.Concerns for cholesterol, fat, and


sugar each had a stronger effect on thepreference for less red meat, comparedwith vitamins and minerals. Themagnitude of effects on the preferencesfor both meatless meals and red meatwas highest for cholesterol concern,followed by sugar and fat concern.Although the data used in this studyshowed an association between redmeat and consumers’ perceived concernabout sugar, it is important to note thatred meat does not contain sugar. It islikely that consumers’ general nutri-tional concern regarding food may notnecessarily reflect their knowledgeabout specific nutrient content.

Conclusion and Implications

A regression analysis was used in thisstudy to analyze socioeconomic andnutritional factors affecting consumerpreferences toward meatless meals andmeals with less red meat. Consumers’preferences for meatless meals andred meat were influenced by a numberof socioeconomic variables such asgeographical location, racial and ethnicbackground, family composition, andhousehold income. There were differ-ences between men and women andmembers of different age groups andat different educational levels. Al-though women were not statisticallydifferent from their male counterpartsregarding their preference for red meat,they generally preferred more meatlessmeals. Nutritional concerns amongconsumers—particularly for choles-terol, fat, and sugar—also played arole in consumers’ preferences. Theseresults have important implications forpublic health and should be consideredwhen developing campaigns related tohealth promotion or meat consumption.

References

Arabmazar, A., & Schmidt, P. (1981). Further evidence of robustness of the Tobitestimator to heteroskedasticity. Journal of Econometrics, 17, 253-258.

Belsley, D.A., Kuh, E., & Welsch, R.E. (1980). Regression Diagnostic, IdentifyingInfluential Data and Source of Collinearity. New York: Wiley.

Blisard, N. (2001). Food spending in American households, 1997/98. U.S.Department of Agriculture, Statistical Bulletin Number 972.

Bureau of Labor Statistics. (2001). Consumer Price Index—Average Price Data.Retrieved from http://data.bls.gov/labjava/outside.jsp?survey=ap.

Frazao, E., & Cleveland, L. (1994). Diet-health awareness about fat andcholesterol—Only a start. FoodReview, 17, 15-22.

Greene, W. (1995). LIMDEP Version 7.0. User’s Manual Reference Guide.Bellport, NY: Econometric Software, Inc.

Gujarati, D.N. (1995). Basic Econometrics. McGraw-Hill, Inc.

Guthrie, J., Fox, J., Cleveland, L., & Welsh, S. (1995). “Who uses nutritionlabeling and what effects does label use have on diet quality?” Journal of NutritionEducation, 27(4), 163-172.

Law, M. (2000). Dietary fat and adult diseases and the implications for childhoodnutrition: An epidemiologic approach. American Journal of Clinical Nutrition,72(5 Suppl), 1291S-1296S.

Lin, C.T.J. (1995). Demographic and socioeconomic influences on the importanceof food safety in food shopping. Agricultural and Resource Economics Review,24 (October) , 190-198.

Long, J.S. (1997). Regression models for categorical and limited dependentvariables. Advanced Quantitative Techniques in the Social Science Series 7.Sage Publications.

Michaud, D.S., Augustsson, K., Rimm, E.B., Stampfer, M.J., Willett, W.C., &Giovannuci, E. (2001). A prospective study on intake of animal products and riskof prostate cancer. Cancer Causes Control, 12(6), 557-567.

Moon, W., Florkowski, W.J., Resurreccion, A.V.A., Beuchat, L.R., Chinnan, M.S.,Raraskova, P., et al. (1998). Consumer concerns about nutritional attributes in atransition economy. Food Policy, 23(5), 357-369.

National Peanut Council. (1997). “Peanut User Study.” A peanut householdsurvey conducted by Gallup Organization, Inc., Princeton, NJ.

Nayga, R.M. Jr. (1996). Determinants of consumers’ use of nutritional informationon food packages. Journal of Agricultural and Applied Economics, 28(2), 303-312.

2002 Vol. 14 No. 2 43

Nayga, R.M. Jr., & Capps, O. Jr. (1994). Analysis of away-from-home and at-home intake of saturated fat and cholesterol. Review of Agricultural Economics,16(3), 387-398.

Nayga, R.M. Jr., & Capps, O. Jr. (1999). US consumers’ perceptions of theimportance of following the US dietary guidelines. Food Policy, 24(5), 553-564.

Powles, J., Hage, B., & Cosgrove, M. (1990). Health-related expenditure patternsin selected migrant groups: Data from the Australian Household ExpenditureSurvey, 1984. Community Health Study, 14(1), 1-7.

Putler, D.S., & Frazao, E. (1994). Consumer awareness of diet-disease relation-ships and dietary behavior: The case of dietary fat. Journal of AgriculturalEconomics Research, 45, 3-17.

Putnam, J., & Allshouse, J.E. (2001). Food consumption, prices, and expenditures,1970-97. Food and Rural Economics Division, Economic Research Service, U.S.Department of Agriculture. Statistical Bulletin No. 965.

Putnam, J., & Gerrior, S. (1997). Americans consuming more grains andvegetables, less saturated fat. FoodReview, 20(3), 2-12.

Sandstead, H.H. (1991). Zinc deficiency. A public health problem? AmericanJournal of Disabled Children, 145(8), 853-859.

Santos, M.L., & Booth, D.A. (1996). Influences on meat avoidance among Britishstudents. Appetite, 27(3), 197-205.

Slattery, M.L., Boucher, K.M., Caan, B.J., Potter, J.D., & Ma, K.N. (1998). Eatingpatterns and risk of colon cancer. American Journal of Epidemiology, 148(1),4-16.

U.S. Department of Agriculture, Economic Research Service. (2001). Per capitafood consumption data system. Retrieved from http://www.ers.usda.gov/data/foodconsumption/.

U.S. Department of Agriculture, National Agricultural Statistics Service. (2001).Retrieved from http://www.usda.gov/nass/pubs/agr01/acro01.htm.

Walker, C. (1995). Meet the new vegetarian. American Demographics, 17(1),9-11.

Winkleby, M.A., Albright, C.L., Howard-Pitney, B., Lin, J., & Fortmann, S.P.(1994). Hispanic/white differences in dietary fat intake among low educated adultsand children. Preventive Medicine, 23(4), 465-473.

2002 Vol. 14 No. 2 44

Educators’ Reports ofFood Acquisition Practices Usedby Limited-Resource Individualsto Maintain Food Sufficiency

Some food acquisition practices of limited-resource individuals were elucidatedthrough interviews with nutrition educators who work regularly with this audience.Practices were characterized as either viable or questionable in terms of theirpotential risks. Practices used to acquire food, or money for food, includedproviding foster care, selling surplus food, switching price tags on food, purchasingfood from private individuals, and seeking out and using road-kill. The foundationof a grounded theory regarding practices used by limited-resource individuals tomaintain food sufficiency was originated. Additional research should verify thesepractices and determine their prevalence among limited-resource audiences,as well as the relative risk associated with using questionable practices. Foodsecurity indicators, nutrition education messages, and policies concerning limited-resource individuals may need to be addressed in light of these findings.

Kathryn M. Kempson, MSDebra Palmer Keenan, PhDPuneeta Sonya Sadani, BSSylvia Ridlen, PhDNancy Scotto Rosato, MA

Rutgers, The State University of New Jersey

ood security has been defined as“access by all people at all timesto enough food for an active,

healthy life [and] includes, at a mini-mum, the ready availability of nutri-tionally adequate and safe foods, andan assured ability to acquire acceptablefoods in socially acceptable ways (e.g.,without resorting to emergency foodsupplies, scavenging, stealing, or othercoping strategies)” (Anderson, 1990).

Food security, hunger, and food in-sufficiency have been broadly studied(Ahluwalia, Dodds, & Baligh, 1998;Alaimo, Briefel, Frongillo, & Olson,1998; U.S. House Select Committeeon Hunger, 1990). Research protocolson food acquisition, however, usuallyrestrict questioning to shoppingpractices used in traditional shoppingvenues (USDA, 1997; U.S. HouseSelect Committee on Hunger, 1990),particular populations such as theelderly, or specific practices such as

food budgeting (Dinkins, 1997;Kendall, Olson, & Frongillo, 1998;Wolfe, Olson, Kendall, & Frongillo,1996). Few studies have explicitlyresearched the practices that limited-resource individuals use to obtainfood and have considered that foodacquisition may occur in nontraditionalvenues (Ahluwalia et al., 1998;Hamelin, Habicht, & Beaudry, 1999;Olson, Rauschenbach, Frongillo, &Kendall, 1997; Petchers, Chow, &Kordisch, 1989). Reporting accuratedata—including the difficulties someAmericans experience in gettingenough food to eat—is critical fornutrition programs and policies(Carlson, Andrews, & Bickel, 1999).

Nutrition educators who work regularlywith limited-resource individuals mayhave strong relationships with theirprogram participants and therefore bekeenly aware of the practices theirclients use to maintain food sufficiency.

F

2002 Vol. 14 No. 2 45

These educators—who may beaccessed with greater ease, less time,and with lower cost—may be a reliablesource for information pertaining tothe limited-resource individuals theyserve. Our research used this alternativesource for information about foodacquisition practices of limited-resource individuals.

Therefore, a study was designed tointerview all EFNEP and FSNEPnutrition educators in New Jersey toidentify the practices that their programparticipants reported using to maintainfood sufficiency. In addition to iden-tifying these practices, our other goalwas to distinguish which practicesposed risks such as those related tofood safety.

Methods

Subject Selection andData CollectionSemi-structured, in-depth interviews,approved by the Rutgers InstitutionalReview Board, were conducted withall 51 EFNEP and FSNEP educatorsthroughout the State. This convenience,nonrandomized sample, representinga single State, consisted of 18 pro-fessionals and 33 paraprofessionalswho had worked for either of theprograms for at least 6 months. Theinterviews were conducted betweenSeptember 1999 and January 2000 bya research team of two faculty membersand two research assistants. Mostinterviews were completed in personand during work hours; three interviewswere conducted via telephone becauseof inclement weather.

During the interviews, educators wereasked to share stories about practicestheir program participants haddiscussed regarding the means theyused to maintain food sufficiency. Tofocus the stories, the interviewers asked

these questions in the order in whichthey are listed:

§ Think back to the classes that youhave taught for EFNEP or FSNEP.What are common things thatpeople have said that they do inorder to get through the month withenough food?

§ What things have people said theydid to get through the month withenough food that surprised you?

§ What things did people do to avoidrunning out of food, that when theyobtained the food, it was unsafe orrisky?

§ What things have people done toavoid running out of food thatseemed to be, or actually were,illegal?

The interviewer used other open-endedquestions to gather additional details,and the interviewer who was notquestioning the educator took extensivenotes. Interviews were audiotaped.However, one educator preferred notto be audiotaped, so interviewer notessubstituted for the transcript. At theend of each interview, educators wereasked about their EFNEP/FSNEP workhistory and personal demographics.

Data AnalysisVerbatim transcripts of the audiotapedinterviews were reviewed for accuracyand compared with the interview notes.Individual food acquisition practiceswere identified and physically cut fromtranscript copies. The four members ofthe research team independently usedthe constant comparative method(Glaser & Strauss, 1967) to organizeinto groups those practices that sharedsimilar strategies of acquiring food. Theresearchers met to compare, contrast,and resolve differences, and then usedthe same approach to organize practicesinto even broader categories that

seemed to depict adequately thecommon themes of food acquisition.In the same manner, practices wereexamined to determine which oneswere viable or questionable.

The researchers defined questionablepractices as those that may have poseda food safety, nutritional, physical,financial, legal, or regulatory risk to theindividuals who used them. To ensurethe validity of these definitions, a fifthresearcher, who was familiar withthe literature in this area, carefullyreviewed all findings and reported anyinconsistencies between the literatureand researchers’ classifications. Dataclassification was performed with thevision that this work would form aportion of the basis of a groundedtheory concerning practices thatlimited-resource individuals use tomaintain food sufficiency (Glaser &Strauss, 1967). In turn, this informationcould be used by nutrition educators totailor messages and by policymakers toexamine practices that put individualsat risk for food insecurity.

Results

The educators were females whoseaverage age was 44.5 and who hadworked in their occupational fields foran average of 7.1 years. Forty-fivepercent were Caucasian; 25 percent,African American; and 19 percent,Hispanic. Nine percent did not specifytheir race/ethnic group.

Two main themes emerged from theinterviews with the educators: (1)practices employed by limited-resourcepeople with the intent of obtaining foodand (2) food management practices(discussed in another paper [Kempson,Keenan, Sadani, Ridlen, & Rosato,2002]). Four categories of foodacquisition practices were identifiedfrom the stories shared by theeducators: (1) Rely on Resources


Offered in the Community, (2) Interactwith Informal Support Systems, (3)Supplement Financial Resources, and(4) Lower Food Costs by UsingShopping Strategies.

Rely on Resources Offeredin the CommunityEFNEP/FSNEP educators reportedthat their clients—limited-resourceindividuals—used three majorstrategies within community systemsto maintain food sufficiency: Clientswere reported to (1) participate infederally funded food programs,(2) attend events to obtain food, and(3) participate in locally sponsoredfood programs (table 1).

Participate in federally funded foodprograms. Federal food programsserved as common venues throughwhich food and money for food wereobtained. Educators reported thatlimited-resource individuals partici-pated in programs such as the FoodStamp Program; Head Start; schoollunch and breakfast programs; andthe Special Supplemental NutritionProgram for Women, Infants, andChildren (WIC) to maintain foodsufficiency.

Attend events primarily to obtain food.Special events at various locations wereattended primarily to take advantage ofthe food that was provided. In thesecases, a church, business, or organi-zation had offered food as an incentiveto increase the turnout at an event, tomarket products, or to cater to attendeesor customers. One educator stated:

“We have a coffee hour betweeneach service . . . , and there area number of folks that come infor our coffee hour [and] do notstay . . . through the churchservice.”

Another commented that although thefood is generally a lot of sweet items

Table 1. Food acquisition practices reported to be used by limited-resourceaudiences to maintain food sufficiency: Rely on resources offered in thecommunity

Practices Possible risk

Participate in Federal Food ProgramsFood StampsHead StartSchool Lunch/Breakfast ProgramsThe Special Supplemental Nutrition Program for Women, Infants, and Children

Attend events primarily to obtain foodChurch fellowshipNutrition education classHappy hour at barsStores offering samples

Participate in locally sponsored food programsFood pantriesLocal programsPrivate businessesNurition sites for seniors and soup kitchensSheltersPrivate individuals Food safety

that were baked and bagels, “it’ssomething to keep [their] stomachfrom growling.”

Educators observed clients attendingan event primarily to obtain food.They noticed a rise in attendance ateducational classes where food/commodities were available. Thispractice was particularly noticeableat the end of the month—presumablywhen funds began to run out. In somecases, class members asked theeducator for food. Happy hours at barsand grocery stores offering sampleswere also visited by limited-resourceindividuals in an attempt to obtain food.

Participate in locally sponsored foodprograms. Food was also obtainedthrough locally subsidized programs—such as church-sponsored food pantries,soup kitchens and shelters, venisonrecovery programs (e.g., Hunters for

the Hungry), Meals on Wheels, andnutrition sites for seniors. Turkeysand other foods were distributed bycharities during the holidays. Privatebusinesses, restaurants, diners, andfast-food establishments sometimesoffered meals throughout the year,hosted holiday dinners, or providedfood in other ways. For example, onerestaurant owner, instead of throwingaway food at the end of the day,prepared plates of food to give to thehungry. In addition, nutrition sites forseniors provided lunches to the elderlyon a regular basis, soup kitchensprovided hot meals, and sheltersprovided both a place to live and asource of food. Private individualsopened community assistance programsor organized neighborhood volunteerfacilities that offered food. Theseprograms could potentially be foodsafety hazards, because the individualsoperating them were unlikely to have

2002 Vol. 14 No. 2 47

Table 2. Food acquisition practices reported to be used by limited-resourceaudiences to maintain food sufficiency: Interact with informal support systems

Practices Risks or possible risks

Exchange resourcesSell surplus food Food safetyTrade forms of public assistance Nutritional; Illegal/regulatory

Manage personal resourcesBudgetEstablish store credit FinancialSystematize payment of bills Financial

Members of support systemAsk for or borrow food or moneyEat at others’ homesShare informationBorrow food stamps Illegal/regulatoryIdentify someone to live with Nutritional; Physical

expertise in safe food handling andwere not required to follow governmentfood safety regulations. This potentiallyplaced the people they served at risk forfoodborne illnesses.

Interact With InformalSupport SystemsPersonal support systems provided anetwork within which limited-resourceindividuals operated to maintain foodsufficiency (table 2). EFNEP/FSNEPeducators identified three majorstrategies their clients used.

Exchange resources. Excess suppliesof specific foods, including rice, cereal,canned and packaged goods, andholiday turkeys, were mentioned by theeducators as being sold for money withwhich other foods could be purchased.Also, public assistance monies, in theform of WIC vouchers for infantformula, were traded for food stamps.These practices not only may havethreatened the infants’ nutritionalstatus, but they also violated programregulations.

Manage personal resources.Budgeting, using credit, and cycling

bill payments were practiced toconserve money for food. Hispaniccommunities reported that owners ofnearby stores (i.e., a “bodega”) oftenestablished an informal credit systemwith familiar customers. Those whopurchased groceries on credit repaidthe storeowner once they received theirfood stamps and/or paychecks, butthese limited-resource customerscontinued to depend on credit for thenext month’s food. In other cases, billpayment was cycled so that the mosturgent bills were paid first; other billswere paid later. As with buyinggroceries on credit, bill cyclingprovided only a short-term solutionfor obtaining food and could befinancially risky if it is used long term.

Use members of support system.Interviews with EFNEP/FSNEPeducators indicated that limited-resource individuals frequentlycollaborated and cooperated withmembers of their support system.They visited friends, neighbors, andfamily members to obtain food ormoney or to be invited to partake ina meal. Information learned fromfriends, relatives, neighbors, health

Budgeting, using credit, andcycling bill payments werepracticed to conserve moneyfor food.


care and education professionals,paraprofessionals, and even in anopportunistic fashion, such as over-hearing a conversation, was used to findresources for food, to increase financialresources, to learn shopping strategies,and to manage household and personalfood supplies better.

Often, food stamps were borrowedfrom others, and those who neededa place to live used neighborhoodconnections to learn of availableresidences and roommates. Many ofthese support systems included casualacquaintances and people involvedin illegal activities, and thus theseaffiliations presented obvious physicalrisks. Theft presented nutritional risks.

Supplement Financial ResourcesEFNEP/FSNEP educators identifiedsix major food acquisition practicesthat their limited-resource clients usedto supplement financial resources andmaintain food sufficiency. The limited-resource clients used strategies toincrease income and decrease expensesto improve their ability to acquire food(table 3).

Increase income through activities.People provided foster care to makeextra money or sold or pawned non-food items to acquire money for food.These non-food items included clothes,donated items, personal possessions,aluminum cans, and electronicequipment.

“People go to pawn shops tosell items that they alreadyhave, like jewelry, watches, . . .appliances, car titles . . . . I’veseen cases where peopleactually lost their cars becausethey didn’t pay back the loanthat they got for the title . . . .They had food, but they losttheir vehicle.”

Table 3. Food acquisition practices reported to be used by limited-resource audiences tomaintain food sufficiency: Supplement financial resources


Increase income through activitiesProvide foster carePawn or sell itemsBegging/panhandling Illegal/regulatoryEarn unreported income Illegal/regulatoryEngage in illegal activities Physical; Illegal/regulatoryGamble Financial

Decrease expenses through activitiesGardenAcquire discarded food Food safetySeek road-kill Food safetyHunt and fish Food safety; Illegal/regulatoryAccess multiple pantries Illegal/regulatory

Relocate to increase incomeCloser to public assistance programsBetter employment opportunities

Relocate to decrease expensesInexpensive housingHousing with shared or unsecured food storage facilities NutritionalAbandoned building Physical; Illegal/regulatory

Use programs to increase incomeObtain Temporary Assistance for Needy FamiliesObtain general assistanceObtain Social Security Income

Use programs to decrease expensesGet subsidized housingParticipate in Self-Help and Resource Exchange (SHARE)

Panhandling was commonly mentionedas a way to increase income for foodacquisition. Babysitting, day work,sharing households with borders whowere not mentioned on applications forpublic assistance, and preparing andselling homemade food were othermeans used to obtain money for food.One limited-resource client boughtseveral cake mixes at the beginning ofthe month, made homemade cupcakes,

and then sold them for profit when herfood allotment for the month wasdepleted. Although an industriouspractice, earning unreported incomeand selling without a license are illegal.

Other illegal activities were alsopursued to increase income. Forexample, meat or non-food items werestolen from stores or individuals andthen sold or pawned. EPNEP/FSNEP

2002 Vol. 14 No. 2 49

educators reported that their clientsstole food from family members duringmealtime visits, from establishmentswhere individuals worked, or in thecase of migrant farm workers, fromcrops.

Adults, whole families, and childrenwere involved in manufacturing,distributing, and selling drugs illegally.

“. . . this started when he wasabout 5 years old and some ofthe older kids in the neighbor-hood would . . . send himaround the corner with onepaper bag. And then he wouldhave to come back withanother paper bag. And theywould . . . give him a couple ofquarters, or whatever, for everytime he came around thecorner. That was how moneywas coming, and that was howhe got food for himself.”

In addition to the legal ramifications,these practices placed individuals atrisk of physical harm.

Legal and illegal gambling, such asbuying lottery tickets or participating insports’ pools, was another way peopleattempted to acquire money for food.

“They thought [that if] they’dspend a dollar here, maybe theycould win a couple million.You [wouldn’t] have to everworry about food again.”

If overused, gambling can lead tofinancial problems that negativelyaffect people’s ability to maintain foodsufficiency.

Decrease expenses through activities.Multiple practices to decrease expenseswere used to reduce food costs, andtherefore improve limited-resourceindividuals’ ability to acquire food.

Gardening was mentioned as a viablemethod; going through dumpstersand picking up discarded food werequestionable methods that placedparticipants at risk for foodborneillnesses.

Hunting and fishing, although oftendone legitimately, were questionablepractices in many cases (e.g., huntingdeer after dark or fishing in contam-inated waters).

“There are quite a few peoplein our area who fish. And theyjust really don’t even try tofind out whether [the water] iscontaminated or safe.”

Sometimes road-kill was sought.EFNEP/FSNEP educators told of somelimited-resource clients who soughtroad-kill and then took it home as asource of meat. When road-kill couldnot be found, it was created.

“. . . just run the animals overwith the car and pick them upand put them on the hood ofthe car and take them home.”

Hunting, fishing, and seeking road-kill,as described, all posed potential foodsafety risks.

Finally, the educators shared storiesof people traveling to multiple pantriesand soup kitchens to obtain the foodthey needed. In many cases, thisviolated the regulations of the foodpantries.

Relocate to increase income anddecrease expenses. Relocation wasused to increase income with whichto purchase food. According to theEFNEP/FSNEP educators, some oftheir limited-resource clients moved tosuburbs or into cities to be closer topublic assistance programs and publictransportation. In other cases, individ-uals or families transferred to less

populated areas, where employmentwas more readily available.

Finding inexpensive housing was apractical way to decrease expenses,which also left more money for food.Living in facilities with common foodstorage areas helped to defray livingexpenses.

“You had a lot of roominghouses, and there would bethree or four people in oneroom. And that would be a wayof getting food, too. Becauseif you did have three or fourpeople in one room . . .everybody was sharing [his orher] food from the food banks,so it would last a little longer.”

Residents of facilities (e.g., someshelters, transitional housing, YMCA’s,and rooming houses) with shared foodstorage areas frequently stole fromeach other, putting individuals at risknutritionally from a lack of sufficientfood. One educator spoke about apersonal experience: “. . . leave a canof food in your room and the next thingyou know, it [would] be gone [becauseof theft].” Residence in abandonedbuildings, although illegal andphysically unsafe, was also reported.

Use programs to increase incomeand decrease expenses. TemporaryAssistance for Needy Families, GeneralAssistance, and Social Security Income(SSI), according to EFNEP/FSNEPeducators, were used by their limited-resource clients to increase financialresources and reduce potential foodinsufficiency. Also, obtaining sub-sidized housing and participating inthe Self-Help and Resource Exchange(SHARE) Program were means used todecrease expenses. Volunteer work is arequired aspect of this program, but theeducators reported that this did notoften happen. Whether increasingincome or decreasing expenses, these


practices served as means throughwhich food sufficiency could bemaintained.

Lower Food Costs by UsingShopping StrategiesAccording to the EFNEP/FSNEPeducators, their limited-resource clientsused three major food acquisitionpractices to maintain food sufficiency.Most of these practices were legal, buta few posed food safety or regulatoryrisks (table 4).

Purchase food from low-cost sources.Discount stores such as wholesale breadoutlets, meat and poultry stores, andproduce outlets offered low-cost food.

“It’s like six cans of peas for abuck. So they buy more staplefoods like canned vegetables,stuff like that so they can haveit on the shelf . . . . It might notbe a piece of meat, but theyhave vegetables.”

Table 4. Food acquisition practices reported to be used by limited-resourceaudiences to maintain food sufficiency: Lower food costs by using shoppingstrategies


Purchase food from low-cost sourcesDiscount storesPrivate individuals and vendors Food safety

Shop for low-cost and value foodsBulk foodsInexpensive foodsCouponsNearly expired foodSale itemsDented and damaged packages Food safetyExpired food Food safety

Engage in illegal shopping practicesShoplift food Illegal/regulatorySwitch price tags on food Illegal/regulatory

Inexpensive food was purchased fromfarmers and neighborhood gardenersand from individuals (unofficial andunregulated “street vendors”) whosold food from unknown sources.Meat was purchased from butcherswho slaughtered animals in theirhomes. The safety of the food could notbe guaranteed in these unregulatedsituations; therefore, individuals mayhave been at risk of contractingfoodborne illnesses.

“We know someone that is abutcher, and he sells the meatcheap. But he keeps it there[unrefrigerated] for a long time.”

Similarly unsafe conditions weredescribed in situations where meat waspurchased from vendors who sold meatfrom their trucks.

“Some guy [comes] aroundwith meat . . . . He is selling itright off his truck . . . . Andthey go there because theyknow it’s very cheap. You can

Regularly attending eventsprimarily to obtain food—suchas church fellowships, nutritioneducation classes, happy hours,and stores offering samples—didresult in obtaining food but isgenerally considered to be asocially unacceptable practiceto acquire food.

2002 Vol. 14 No. 2 51

get a big chunk of steak for $3.That would cost you about $15or $20 in the store. There’s noreal refrigeration to keep themeat frozen . . . . It’s just alittle portable refrigerator thatlooks like it could burn out atany time. On a hot summerday, that’s not good. He’sdriving around all day sellingmeat through the city.”

Shop for limited-cost and value foods.Limited-resource clients, according toEFNEP/FSNEP educators, use somecost-cutting strategies to save moneywhile shopping. Food items werepurchased in large quantities to receivebulk discounts; in some cases, this foodwas shared with friends or families.Expensive foods, such as fresh fruits,were avoided in favor of their lowerpriced canned or frozen counterparts.Inexpensive foods also were used tolower the overall food bill. Examplesof these foods included Ramen-stylenoodles, stews, hamburger, macaroniand cheese, canned foods, bones tomake soup, generic or store brands,flour, dry foods, rice, tuna, peanutbutter, and pasta. Using coupons wasoccasionally mentioned as a way ofsaving money on food purchases.“Almost” expired foods were alsopurchased. Examples of such purchaseswere day-old bread, cheese, meat, andproduce.

Practices that put individuals at riskin terms of food safety were also usedin the quest to obtain low-cost food.Multiple educators reported stories ofpeople purchasing dented cans becausethe prices were reduced or shoppingin stores that specialized in the distri-bution of such items.

“Actually, all of the super-markets also have [dented cans]. . . usually in the back of thestore . . . . Some cans don’teven have labels on them.”

Stores were also reported as havingsold damaged, expired, or improperlyrefrigerated foods. Additionally,expired foods with reduced prices weresought and purchased by the limited-resource individuals.

Engage in illegal shopping practices.Foods, such as grapes, were eaten whileshopping in grocery stores, and avariety of practices were used to takefood from the grocery stores and/orsupermarkets. EFNEP/FSNEPeducators reported that some of theirlimited-resource clients engaged in thefollowing practices:

“They would shoplift. If theywere pregnant, they felt that itwas easier to hide the food . . .in their pants.”

“They take the bag into thestore, whatever store bag thatbelongs to the store, and theywill take it in the store andactually go shopping.”

Price tags were also switched onfoods; thus, expensive food could bepurchased for less money. Leavingrestaurants without paying for themeal was mentioned as well.

Discussion

Our findings showed that foodacquisition methods of limited-resource individuals had not beenfully documented when staff fromthe New Jersey Expanded Food andNutrition Education Program (EFNEP)and Food Stamp Nutrition EducationProgram (FSNEP) anecdotallydescribed novel practices. Thesestories included purchasing meatthat was butchered by a neighbor ina basement and fish that was caughtfrom contaminated waters, baggingperishable leftovers from dinners atcommunity centers and then not

refrigerating them for many hours,and claiming paternity for unrelatedchildren to qualify for public assistanceand food stamps. These anecdotescharacterized facets of food insecuritythat have been minimally addressedwithin the food security community.

One study limitation that should berecognized is that the data wereobtained from a second-hand con-venience, nonrandomized sample.Accuracy of responses supplied by theeducators has yet to be validated withresponses received first hand fromlimited-resource individuals. However,the nutrition educators had strongrelationships with their programparticipants and were acutely awareof the practices their clients usedto maintain food sufficiency. It isuncertain whether limited-resourceindividuals would feel comfortablesharing their stories with unfamiliarresearchers. In fact, because of thefear of possible punitive outcomes,it is possible that individuals wouldnot report questionable practices toresearchers. Therefore, we proposethat the data gleaned from interviewingnutrition educators are likely morecomplete than data gathered directlyfrom limited-resource individuals.Nevertheless, the results of this studymust be verified with limited-resourceindividuals.

Extrapolation of the prevalence of eachpractice into a larger population is notvalid. We collected qualitative data.If our results are quantified, seriousinterpretation errors could result. Forexample, it is possible that a numberof nutrition educators were aware ofthe same limited-resource individual’sunique food acquisition practice.In that case, the quantification ofresponses could greatly overestimatethe prevalence of that particularpractice. The determination of theprevalence of these practices is an areafor future research to be conducted, in


the form of a survey, directly withlimited-resource individuals. Bothviable and questionable practicesidentified in this research confirmedand expanded upon previous findings.The significant findings regarding useof nontraditional sources to acquirefood and the sharing of informationwere documented, and a portion of thebasis of a grounded theory concerningthe food acquisition practices oflimited-resource individuals wasformed (Glaser & Strauss, 1967):

• Previously identified copingstrategies related to foodacquisition were more clearlydefined and elaborated.

• New coping strategies related tofood acquisition were revealed.

• Emergent findings, concept-ualization, and categorizationhave transpired, yet relationshipsamong emergent findings needto be defined.

This study and others show that peoplewill resort to many, and sometimesdrastic and often illegal, means toensure that they and their familiescan eat. Begging, earning unreportedincome, gambling, selling personalpossessions, obtaining discarded food,and trying to get sent to jail for ensuredaccess to food were verified previouslyin the literature (Ahluwalia, 1998;Austin, 1996; Curtis, 1995; Eisinger,1996; Hamelin, 1999; Herth, 1996;Kendall, 1996; Morton, 1997; Olson,1997; Petchers, 1989; Pflugh, 1999;Quandt, 2000; U.S. House SelectCommittee on Hunger, 1990). Engagingin illegal activities has also beendetermined as a means of earning foodmoney (Petchers, Chow, & Kordisch,1989). Hamelin, Habicht, and Beaudry(1999) provided a possible rationale forthese kinds of activities: “Eventually,the search for food takes precedenceover previously held values.”

Many practices described in this studywere not identified in the scientificliterature. The finding that peopleprovide foster care primarily for thepurpose of earning extra money forfood requires further study to ensurethe well-being of children in thesecases. Regularly attending eventsprimarily to obtain food—such aschurch fellowships, nutrition educationclasses, happy hours, and storesoffering samples—did result in obtain-ing food but is generally considered tobe a socially unacceptable practiceto acquire food. Private individualssponsoring food programs, sellingsurplus food, trading WIC checks forfood stamps, establishing store credit,switching price tags on food, and pur-chasing food from private individualswere also novel findings. Although thepractice of searching for or using road-kill for food has not been elucidatedin the scientific literature, multiplenewspaper stories have reportedanecdotal evidence of this practice(Firestone, 1999; Stuever, 1999).

Even under harsh conditions, somepeople were reportedly able to maintainfood sufficiency, while others were not.The ways in which people interactedwith their support networks enhancedtheir abilities to improve food suffi-ciency in a variety of ways. In fact,many of the practices reported in ourstudy require learning through informalinteractions. The sharing of infor-mation, an overriding theme in thisstudy, occurred in every category offood acquisition practice identified.People learned from others about thelocation of assistance programs, ideasfor increasing income and savingmoney, as well as ways to obtain food.Dependence on members in supportsystems—by asking for or borrowingmoney, using food stamps, eatingwith other people, identifying moreeconomical or convenient places tolive, and sharing information—has beendocumented previously in the literature

(Ahluwalia et al., 1998; Curtis &McClellan, 1995; Hamelin et al., 1999;Kendall et al., 1996; Olson et al., 1997;Petchers et al., 1989). One educatorstated,

Well, I think I’ve seen thespectrum of very talented, veryskillful individuals . . . whohave learned how to navigatethe system and use the systemto their advantage. And thenI’ve seen those who are ill-informed . . . in terms of howto stretch their dollar . . . ornavigate the system. And whenI talk about ‘navigate thesystem’ we have all kinds ofsupport systems across thiscountry so that no one will gohungry . . . so if you know howto manage, navigate, you canprovide for your family.

Although not the focus of this study,pride and embarrassment werementioned as barriers to obtainingassistance. These obstacles have beendocumented and elucidated by others(Kendall et al., 1998; Petchers et al.,1989).

Although discount shopping wasreported, some educators indicated thatmany people still shop at nearby, moreexpensive convenience stores. Foodpurchased at convenience stores isremarkably more expensive than foodfrom supermarkets and large grocerystores (Mantovani, Daft, Macaluso, &Hoffman, 1997). Despite the conven-tional wisdom that it is unwise forlimited-resource individuals to shopat expensive venues, some people hadno other choice because of a lack oftransportation. That a lack of trans-portation translates to limited accessto less expensive stores has beenconfirmed by many studies (Dinkins,1997; Kendall et al., 1996; Quandt,McDonald, Arcury, Bell, & Vitolins,2000; Travers, 1996; U.S. HouseSelect Committee on Hunger, 1990).

2002 Vol. 14 No. 2 53

Many of the food acquisition practicesreported in our study involved risks tothe individuals who resorted to usingthem. Food acquisition occurredoutside the traditional shopping venuesthat are closely regulated for foodsafety (e.g., supermarkets andrestaurants), thus placing individualsat risk of foodborne illnesses. Multipleother practices increased food safetyrisks, as well as risk for nutritionaldeficiencies, impaired physical well-being, the dangers of arrest, removalfrom beneficial programs, or theperpetuation of a cycle of financialdependence on credit. In cases whereindividuals were food insufficient,survival took clear precedence overthese risks.

It is plausible that food securityindicators currently in common use donot incorporate the issues of food safetyand social acceptability, which are partsof the definition for food security. Thisstudy’s findings lead the authors tosuggest the possibility that some peoplewho have learned to function wellwithin their environments would reportthat they are food sufficient and wouldbe classified as food secure by mostindicators. However, these individualswould, in fact, be food insecure becauseof their use of unsafe or socially un-acceptable food acquisition practicessuch as: seeking out road-kill, fishingfrom contaminated waters, andacquiring discarded food.

Part of the definition of food securityby the Life Sciences Research Office is“to acquire acceptable foods in sociallyacceptable ways” (Anderson, 1990);this implies food acquisition “withoutresorting to emergency food supplies,scavenging, stealing and other copingstrategies” (Hamelin et al., 1999).Individuals who are apparently foodsufficient may be food insecure becausethe practices they use are unsustainable(cutting on quantity and/or quality offood, buying food on credit)” (Hamelinet al., 1999). Food sufficiency simply

implies an adequate amount of food anddoes not address whether acceptablefood was obtained in a sustainable,socially acceptable manner.

Nonetheless, prior to including theseconstructs in indicators of foodsecurity, research needs to be con-ducted with the target population toverify these data, to determine therelative risks of engaging in each of thequestionable food acquisition practicesreported, and to assess the prevalent useof these practices. Use of certain foodacquisition practices can indicate agreater degree of food insecurity. Forexample, shopping economically andusing coupons are acceptable, risk-freepractices; deliberately committing acrime to be sent to prison to secure afood supply and searching a dumpsterfor food are not. It would also behelpful to know which food acquisitionpractices are used regularly by thegeneral population and which areunique to limited-resource individuals,and more specifically, those living invarious areas, such as more urban areas.

Nutrition education messages mayneed to be reviewed in light of thesefindings. For example, as a result oflearning that the purchase of dentedcans is a very common practice, theNew Jersey EFNEP and FSNEPprograms changed the message tolimited-resource audiences from “donot buy dented cans” to “buy premiumdented cans.” Premium dented cansare those that are not bulging, notpunctured, and not dented along theprimary and secondary seams. In thecase of transportation difficulties, itmay be more helpful to teach peoplehow to shop economically at conven-ience stores. Other practical educationaltopics that may need to be addressedinclude the safety of nonperishablefood and budgeting issues. Broadlyaccessed programs, such as EFNEP,FSNEP, and WIC are viable mech-anisms for information disseminationabout food acquisition practices.

. . . shopping economically andusing coupons are acceptable,risk-free practices; deliberatelycommitting a crime to be sent toprison to secure a food supplyand searching a dumpster forfood are not.


Finally, policy implications andmeasurement issues may need to beaddressed in light of findings thatsome limited-resource individualswho successfully navigate theirenvironments to maintain foodsufficiency may have been assessedas food secure, when in fact they maybe food insecure.

When planning educational programs,making policy decisions, and devel-oping policy and program guidelines,educators and policymakers mustunderstand the mindsets and situationsof those who will be affected. In thisresearch, it was clear that educatorsand policymakers whose work involveslimited-resource audiences shouldalways keep this in mind:

“[Some] need to eat to survive,so you know whatever foodsthat are available they’re goingto eat . . . . They’re trying justto survive.”

Acknowledgments

Funding was provided by the USDAEconomic Research Service grantnumber 43-3AEM-8-8052 to U.C.Davis’ USDA Small Grant Projects,which was then subcontracted underResearch Agreement number K981834-05. The NJ Agricultural ExperimentStation also provided funding underproject number 14170. The authorsacknowledge the New Jersey ExpandedFood and Nutrition Education Programand Food Stamp Nutrition EducationProgram faculty, program associates,and community assistants who sharedtheir stories; Sara Fein for her insightfulreview of our research; and RayaneAbuSabha for her helpful editorialcomments.

References

Ahluwalia, I.B., Dodds, J.M., & Baligh, M. (1998). Social support and copingbehaviors of low-income families experiencing food insufficiency in NorthCarolina. Health Education & Behavior, 25(5), 599-612.

Alaimo, K., Briefel, R.R., Frongillo, E.A. Jr., & Olson, C.M. (1998). Foodinsufficiency exists in the United States: Results from the Third National Healthand Nutrition Examination Survey (NHANES III). American Journal of PublicHealth, 88(3), 419-426.

Anderson, S.A. (Ed.). (1990). Core indicator of nutritional state for difficult-to-sample populations. Journal of Nutrition, 120, 1559-1600.

Austin, C.K., Goodman, D.E., & Van Halderen, L.L. (1996). Absence ofmalnutrition in a population of homeless veterans. Journal of the AmericanDietetic Association, 96(12), 1283-1285.

Bauman, K.J. (1999). Extended measures of well-being: Meeting basic needs.Current Population Reports. U.S. Census Bureau, U.S. Department of Commerce,Economics and Statistics Administration.

Carlson, S.J., Andrews, M.S., & Bickel, G.W. (1999). Measuring food insecurityand hunger in the United States: Development of a national benchmark measureand prevalence estimates. Journal of Nutrition, 129 (suppl 2), 510S-516S.

Curtis, K.A., & McClellan, S. (1995). Falling through the safety net: Poverty, foodassistance and shopping constraints in an American city. Urban Anthropology,24(1-2), 93-135.

Dinkins, J.M. (1997). Food preparers: Their food budgeting, cost-cutting, and mealplanning practices. Family Economics and Nutrition Review, 10(2), 34-37.

Eisinger, P. (1996). Toward a national hunger count. Social Service Review, 70(2),214-234.

Firestone, D. (1999, March 14). A road-kill proposal is food for jokesters. New YorkTimes, p. 20.

Glaser, B.G., & Strauss. A.L. (1967). The Discovery of Grounded Theory. Chicago:Aldine Publishing Company. pp.101-115.

Hamelin, A.-M., Habicht, J.-P., & Beaudry, M. (1999). Food insecurity:Consequences for the household and broader social implications. The Journal ofNutrition, 129(2s), S525-S528.

Herth, K. (1996). Hope from the perspective of homeless families . Journal ofAdvanced Nursing, 24(4), 743-753.

Kempson, K.M., Keenan, D.P., Sadani, P.S., Ridlen, S., & Rosato, N.S. (2002).Food management practices used by limited-resource individuals in New Jersey tomaintain food sufficiency. Journal of the American Dietetic Association, 102(12),1795-1799.

2002 Vol. 14 No. 2 55

Kendall, A., Olson, C.M., & Frongillo, E.A. Jr. (1996). Relationship of hunger andfood insecurity to food availability and consumption. Journal of the AmericanDietetic Association, 96, 1019-1024.

Kendall, A., Olson, C.M., & Frongillo, E.A. Jr. (1998). Hunger and food insecurity inthe elderly. Journal of Aging and Health, 10, 327-350.

Mantovani, R.E., Daft, L., Macaluso, T.F., & Hoffman, K. (1997). Food retailers inthe food stamp program: Characteristics and service to program participants.Alexandria: UDSA Food and Consumer Service.

Morton, J.F., & Guthrie, J.F. (1997). Diet-related knowledge, attitudes, and practicesof low-income individuals with children in the household. Family Economics andNutrition Review, 10(1), 2-15.

Olson, C.M. (1999). Nutrition and health outcomes associated with food insecurityand hunger. Journal of Nutrition, 129(2s), 521S-524S.

Olson, C.M., Rauschenbach, B.S., Frongillo, E.A. Jr., & Kendall, A. (1997). Factorscontributing to household food insecurity in a rural upstate New York county. FamilyEconomics and Nutrition Review, 10, 2-17.

Petchers, M.K., Chow, J., & Kordisch, K. (1989). Urban emergency food centerclients: Characteristics, coping strategies and needs. Journal of Sociology & SocialWelfare, 16(2), 195-203.

Pflugh, K.K., Lurig, L., Von Hagen, L.A., Von Hagen, A., & Burger, J. (1999).Urban angler’s perception of risk from contaminated fish. Science of the TotalEnvironment, 228(2-3), 203-218.

Quandt, S.A., McDonald, J., Arcury, T.A., Bell, R.A., & Vitolins, M.Z. (2000).Nutritional self-management of elderly widows in rural communities. TheGerontologist, 40(1), 86-96.

Stuever, H. (1999, December 1). Bless this food; In the aftermath of a deer accident,a suburban butcher makes sense (and some venison) of a random world . WashingtonPost, p. C01.

Travers, K.D. (1996). The social organization of nutritional inequities. Social Scienceand Medicine, 43(4), 543-553.

U.S. Department of Agriculture. (1997). Understanding the Food Choices of Low-income Families. Retrieved from www.fns.usda.gov/oane/menu/Published/NutritionEducation/Files/NUTRI.PDF.

U.S. House Select Committee on Hunger. (1990). Obtaining food: Shoppingconstraints of the poor. The Black Scholar, (Jan, Feb, Mar), 6-16.

Wolfe, W.S., Olson, C.M., Kendall, A., & Frongillo, E.A. Jr. (1996). Understandingfood insecurity in the elderly: A conceptual framework. Journal of NutritionEducation, 28(3), 92-100.


Trends in Food and NutrientIntakes by Children in theUnited States

Cecilia Wilkinson Enns, MS, RDSharon J. Mickle, BSJoseph D. Goldman, MA

U.S. Department of AgricultureAgricultural Research Service

D

Monitoring dietary trends can make it easier to target dietary guidance. Trendsin intakes among children age 6 to 11 years were examined by using data fromthe Continuing Survey of Food Intakes by Individuals (CSFII) 1994-96, 1998; theCSFII 1989-91; and the Nationwide Food Consumption Survey 1977-78. Increaseswere seen in intakes of soft drinks, total grain products, grain mixtures, crackers/popcorn/pretzels/corn chips, fried potatoes, noncitrus juices/nectars, lowfat milk,skim milk, cheese, candy, and fruit drinks/ades. Decreases in intake were ob-served in whole milk and total milk, yeast breads/rolls, green beans, corn/greenpeas/lima beans, beef, pork, and eggs. Lower percentages of calories from fatwere partly due to increased carbohydrate intakes. Children had decreasesin vitamin B12 and increases in thiamin and iron. Servings per day from the foodgroups of the Food Guide Pyramid were used to discuss diet quality during 1994-96, 1998. For any given Pyramid group, less than one-half of the children con-sumed the recommended number of servings, and their intakes of discretionaryfat and added sugars were much higher than recommended. Guidance shouldcontinue to encourage increases in intakes of whole grains, fruits, dark-green anddeep-yellow vegetables, legumes, nonfat or lowfat dairy products, and lean meats;decreases in fats and added sugars; and increases in activity. Effective nutritioneducation efforts for children should be supported at every level.

ietary guidance in the UnitedStates is given in terms of thetypes and amounts of food

people should try to eat in a day (U.S.Department of Agriculture [USDA],1996). To provide nutrition educationmessages that help people eat morehealthfully, we need to know whatpeople are actually eating, whetherfood intakes are changing, and, if so,how they are changing. The mostrecent USDA survey of dietary intakesprovides an opportunity to update ourknowledge of trends in dietary intakesby children.

To see whether children’s food intakeshave changed over time, we compared

nationally representative estimates fromthe most recent USDA survey of dietaryintakes with similar estimates from twoprevious USDA surveys. The threesurveys were the Nationwide FoodConsumption Survey (NFCS) 1977-78,the Continuing Survey of Food Intakesby Individuals (CSFII) 1989-91, andthe CSFII 1994-96, 1998 (Tippett et al.,1995; USDA, 1983; USDA, 1999;USDA, 2000a). The estimates reportedin this study are of food intakes, thepercentages of individuals consumingfoods, and nutrient intakes for girls andboys age 6 to 11 years during all threeperiods.

2002 Vol. 14 No. 2 57

Design and Methods

The Three SurveysThe CSFII 1998 was conductedbecause a larger sample of childrenunder 10 years old was needed foradequate estimation of dietary exposureto pesticide residues (USDA, 2000a,documentation section 2). The CSFII1998 was designed to be merged withthe CSFII 1994-96, and combineddata have been released (USDA,1999; USDA, 2000a). Data collectionmethods, instruments, and procedureswere the same in 1998 as in 1994-96(Tippett & Cypel, 1998; USDA,2000a). In each of the 4 years of theCSFII 1994-96, 1998, a nationallyrepresentative sample of noninstitu-tionalized individuals residing in theUnited States was surveyed. However,the CSFII 1998 sample included onlychildren under 10 years old. Thesampling weights developed for thecombined 4-year data set permit the3-year and 1998 samples to be com-bined statistically and ensure that thecombined sample is appropriatelyrepresentative of the U.S. population.

The CSFII 1994-96, 1998 was the mostrecent study in the evolving series ofUSDA food and nutrient intake surveysthat also includes the two earliersurveys (Tippett, Enns, & Moshfegh,2000). Differences among the threesurveys in sampling and methodologyare discussed briefly in the followingparagraphs. More information onmethods in the NFCS 1977-78 and theCSFII 1989-91 is available elsewhere(Tippett et al., 1995; USDA, 1983).

The target population covered all 50States in 1994-96, 1998 versus the 48conterminous States in 1977-78 and1989-91. In 1989-91 and 1994-96,1998, the low-income population wasoversampled. In 1977-78 and 1989-91,all children in the sample householdswere eligible for inclusion in the

survey; in 1994-96, 1998, selectedindividuals within each householdwere eligible. The number of childrenage 6 to 11 and the all-individualsDay-1 response rate, respectively, foreach survey are 4,107 and 56.9 percent(NFCS 1977-78); 1,476 and 57.6percent (CSFII 1989-91); and 2,000and 81.5 percent (CSFII 1994-96,1998).

In 1977-78 and 1989-91, dietary datawere collected on 3 consecutive daysby using a 1-day dietary recall and a2-day dietary record. In 1994-96, 1998,the number of days was reduced totwo, partly to reduce respondent burden(Tippett & Cypel, 1998). Both days ofCSFII 1994-96, 1998 dietary data werecollected by means of 1-day dietaryrecalls; interviews were on nonconsecu-tive days, 3 to 10 days apart, to ensurethat nutrient intakes on the 2 dayswould be statistically uncorrelated.

Between the earlier surveys and theCSFII 1994-96, 1998, the 1-day recallwas modified to include multiplepasses through the list of all foods andbeverages recalled by the respondent(Tippett & Cypel, 1998). In 1977-78and 1989-91, an adult householdmember reported intake informationfor children under 12 years old (Tippettet al., 1995; USDA, 1983). In 1994-96,1998, children 6 to 11 years old wereasked to describe their own food intakeand were assisted by an adult householdmember. When necessary, additionalinformation was obtained from schoolcafeteria personnel or baby-sitters.All these measures were designed toimprove the completeness of the datacollected.

The USDA Survey Nutrient Databasewas updated on an ongoing basis toincorporate additional nutrients andimproved nutrient values as well as toreflect changes in foods on the market(Tippett & Cypel, 1998; Tippett et al.,1995; USDA, 1987; USDA, 1993).

Presentation of EstimatesBecause the number of survey daysand the method of data collection onDay 2 differed among the surveys,tables comparing food and nutrientintake estimates among the surveysare based on only the first day’s datacollected from each individual. Usingthese data maximizes comparabilityamong surveys. One-day data areappropriate for comparisons of groupmeans. All estimates are weighted tobe nationally representative.

Mean food intakes are presented“per individual,” meaning they includeintakes by both consumers and non-consumers of the food group. Tocalculate “per user” intakes of foods,researchers may divide the mean intakeof a food group by the percentage ofindividuals using that food group,expressed as a decimal. Because onlyselected food subgroups are presented,subgroup intakes will not sum to thefood group total.1 Food mixtures werenot broken down; mixed foods reportedby respondents were grouped by theirmain ingredient.2 One effect of thismethod of classifying foods is theinflation of some food groups orsubgroups (e.g., meat mixtures) anddeflation of others (e.g., sugars andsweets) relative to the amounts theywould contain if all ingredients weredisaggregated.

Estimates based on a small number ofobservations or on highly variable datamay tend to be less statistically reliablethan estimates based on larger samplesizes or on less variable data. Standarderrors may be used to calculate ameasure of the relative variability of

1Readers interested in subgroups not includedhere are directed to Tippett et al., 1995; USDA,1983; and USDA, 1999.

2See “Table Notes” in Tippett et al., 1995, andUSDA, 1983; see “Descriptions of FoodGroups” in USDA, 1999.


an estimate called the coefficient ofvariation, the ratio of the standard errorto the estimate itself. Because the CSFIIhas a complex sample design, samplingweights and specialized standard errorestimation procedures were used incomputing the estimates and standarderrors (USDA, 2000a, documentationsection 5). SAS version 8.2 (1999)and SUDAAN version 7.5.1 (Shah,Barnwell, & Bieler, 1997) were usedfor statistical calculations.

In the tables, we flagged estimatesthat are potentially less reliable becauseof factors such as small sample sizesor large coefficients of variation. Theguidelines that were used for determin-ing when a statistic may be less reliableinvolve the use of a variance inflationfactor in the role of a broadly calculateddesign effect; those guidelines havebeen described in detail elsewhere(USDA, 1999, appendix B). Thevariance inflation factors used in thisstudy are 1.19 (1977-78), 2.26(1989-91), and 2.24 (1994-96, 1998).

Approximate t-tests were performedto determine whether food and nutrientintakes and the percentages of individu-als using foods were significantlyhigher or lower in 1977-78 versus1989-91; 1989-91 versus 1994-96,1998; and 1977-78 versus 1994-96,1998. All told, 460 pairs of estimateswere compared. Because the analysisinvolved such a large number ofcomparisons, we used conservativecriteria for significance. When signifi-cant differences are discussed in thetext, they may be referred to either as“changes” (or values may be said tohave risen/fallen or to be higher/lowerin 1994-96, 1998 than in 1977-78) oras “trends.”

The term “change” is used only ifintakes (or percentages using) in 1977-78 and 1994-96, 1998 were differentwhen p was less than 0.001. The term“trend” is used only if two criteria were

met: (1) mean intakes (or percentagesusing) either rose or fell progressivelyfrom one survey to the next (e.g., intakeX rose between 1977-78 and 1989-91,then rose again between 1989-91 and1994-96, 1998), and (2) p was less than0.05 for both comparisons. For eachtrend, the level of significance noted inthe tables (< 0.05 or < 0.01) is the onethat is true of both the 1977-78 versus1989-91 t-test and the 1989-91 versus1994-96, 1998 t-test. For example, ifthe 1977-78 versus 1989-91 t-test wassignificant at p < 0.01 but the 1989-91versus 1994-96, 1998 t-test wassignificant at p < 0.05, the latterlevel is shown in the table.

Results and Discussion

BeveragesIn the past 20 years, the overall pictureof beverage intakes by children haschanged considerably. There was adecreasing trend in intake of total fluidmilk—driven by a reduction in theintake of whole milk—for girls age 6to 11 and a parallel change for boys thesame age; an increasing trend in intakesof soft drinks was seen for both girlsand boys (tables 1 and 2). In 1977-78children drank about four times asmuch fluid milk as any other beverage;by 1994-96, 1998 they drank onlyabout 1.5 times as much milk as softdrinks. Other beverages also contrib-uted to the shift to a lesser degree. Bothintakes and percentages of childrenusing beverages from the groups“noncitrus juices and nectars” and“fruit drinks and ades” were higher in1994-96, 1998 than in 1977-78 (tables1 through 4).

The shift in beverage intakes is ofnutritional concern. Guenther (1986)found negative associations betweenintake of soft drinks and intakes ofmilk, calcium, magnesium, riboflavin,vitamin A, and vitamin C. Harnack,

In 1977-78 children drank aboutfour times as much fluid milk asany other beverage; by 1994-96,1998 they drank only about 1.5times as much milk as softdrinks.

2002 Vol. 14 No. 2 59

Table 1. Trends and changes in girls’ (6 to 11 years) mean intakes from selectedfood groups

Intake (grams)Food group 1977-78 1989-91 1994-96, 1998 Change1 Trend2

Grain products 230 268 280 +50Yeast breads and rolls 51 46 43 -8Ready-to-eat cereals 18 21 20Cakes, cookies, pastries, pies 34 33 42Crackers, popcorn, pretzels, corn chips 5 9 14 +9 **Mixtures mainly grain 63 93 101 +38

Vegetables 159 128 116 -42White potatoes 56 55 46 Fried white potatoes 17 26 25 +9Dark-green vegetables 7 4 5Deep-yellow vegetables 7 73 4Tomatoes 14 14 15Green beans 10 7 5 -5Corn, green peas, lima beans 22 16 12 -10

Fruits 159 194 169Citrus juices 60 55 54Apples 27 32 21Melons and berries 7 63 8Noncitrus juices and nectars 14 44 42 +28

Milk and milk products 492 430 382 -110 * Fluid milk 417 339 283 -133 **

Whole milk 244 171 108 -136 ** Lowfat milk 64 146 136 +72 Skim milk 14 15 29

Milk desserts 28 31 3 Cheese 8 12 14 +6Meat, poultry, and fish 157 141 130 -28

Beef 40 18 18 -22Pork 13 9 5 -8 *Frankfurters, sausages, luncheon meats 17 20 19Chicken 18 17 17Fish and shellfish 6 93 5Mixtures mainly meat, poultry, fish 55 63 60

Eggs 16 15 11 -5Legumes 20 21 12Fats and oils 9 9 7Sugars and sweets 28 41 41 +13

Candy 5 8 12 +7Beverages 250 264 370 +120

Tea 51 40 34Fruit drinks and ades 91 86 134 +43Carbonated soft drinks 106 136 200 +94 *

1Change = mean intakes in 1977-78 and 1994-96, 1998 are significantly different at p < 0.001.2Trend = mean intake rose or fell progressively from 1977-78 through 1989-91 to 1994-96, 1998.3Estimate is based on small sample size or coefficient of variation > 30 percent.* = trend significant at p < 0.05.** = trend significant at p < 0.01.

Stang, and Story (1999) reported apositive association between consump-tion of nondiet soft drinks and energyintake in an analysis of CSFII 1994data. Wyshak (2000) found that high-school-age girls who drink carbonatedbeverages may have a higher risk ofbone fractures than do girls who donot drink carbonated beverages. In a19-month-long prospective study,Ludwig, Peterson, and Gortmaker(2001) observed an association betweenconsumption of sugar-sweetened drinksand childhood obesity. Because thestudies by Guenther (1986), Harnack etal. (1999), Wyshak (2000), and Ludwiget al. (2001) were observational, itcannot be inferred that the relationshipsbetween soft drinks and the negativeoutcomes described were causal.Further research is needed in this area.

FoodsOverall, the intakes of grain productswere one-fifth to one-third higher in1994-96, 1998 than in 1977-78 for girlsand boys age 6 to 11 (tables 1 and 2).In all three surveys, the subgroup“mixtures mainly grain”—grain-basedmixtures such as pasta with sauce, ricedishes, and pizza—accounted for thelargest share (by weight) of grainproducts eaten by children. Intakesof the grain mixtures subgroup and thepercentages of children using grainmixtures increased between 1977-78and 1994-96, 1998 (tables 3 and 4).

Increasing trends were observed inchildren’s intakes of grain-basedsnack foods from the group “crackers,popcorn, pretzels, and corn chips.” Atrend toward higher intakes of ready-to-eat cereals was also evident forboys. Both intakes and percentages ofindividuals using yeast breads and rollswere lower for girls and boys in 1994-96, 1998 than in 1977-78 but did notmeet the definition of a trend. Yeastbreads and rolls are common compo-nents in sandwiches, and some sand-wiches (especially fast-food items) are


Table 2. Trends and changes in boys’ (6 to 11 years) mean intakes from selectedfood groups

Intake (grams)Food group 1977-78 1989-91 1994-96, 1998 Change1 Trend2

Grain products 244 296 318 +74Yeast breads and rolls 57 47 46 -11Ready-to-eat cereals 20 24 31 +11 *Cakes, cookies, pastries, pies 40 37 40Crackers, popcorn, pretzels, corn chips 5 9 15 +11 *Mixtures mainly grain 61 108 115 +53

Vegetables 154 130 115 -39White potatoes 61 57 50 Fried white potatoes 18 26 27 +8Dark-green vegetables 6 5 5Deep-yellow vegetables 7 4 5Tomatoes 11 12 16Green beans 8 7 5 -4Corn, green peas, lima beans 23 14 11 -11


Milk and milk products 527 459 450 -77Fluid milk 447 374 335 -112 Whole milk 256 155 121 -135 Lowfat milk 70 193 172 +102 Skim milk 15 15 33Milk desserts 30 24 35Cheese 7 10 12 +5

Meat, poultry, and fish 172 145 154Beef 39 18 19 -21Pork 15 11 7 -8Frankfurters, sausages, luncheon meats 19 18 24Chicken 21 18 20Fish and shellfish 6 10 6Mixtures mainly meat, poultry, fish 60 64 72

Eggs 19 15 12 -6Legumes 22 13 13Fats and oils 10 9 7Sugars and sweets 30 28 42 +13

Candy 5 9 12 +8Beverages 264 329 413 +149 **


1Change = mean intakes in 1977-78 and 1994-96, 1998 are significantly different at p < 0.001.2Trend = mean intake rose or fell progressively from 1977-78 through 1989-91 to 1994-96, 1998.* = trend significant at p < 0.05.** = trend significant at p < 0.01.

In 1994-96, 1998, only 24 percentof girls and 23 percent of boysconsumed the number ofservings of fruit recommendedin the Food Guide Pyramid basedon their caloric intake.

2002 Vol. 14 No. 2 61

categorized under “mixtures mainlymeat, poultry, fish”; intake estimates foryeast breads and rolls would be higherif the breads and rolls from thosesandwiches were included here.

In 1994-96, 1998, only 39 percentof girls and 47 percent of boys con-sumed the number of servings of grainproducts recommended in the FoodGuide Pyramid based on their caloricintake (USDA, 2000b). DespitePyramid recommendations to choose“several servings a day” of whole-grainfoods (USDA, 1996), intake of wholegrains by children in 1994-96, 1998was only 1 serving per day or less.

Children’s intakes from the category“total vegetables” were lower in 1994-96, 1998 than in 1977-78. It is impor-tant to remember that vegetables arefrequently consumed as part of meatmixtures and grain mixtures; in intakesby adults in the CSFII 1994, vegetablesaccounted for about 24 and 28 percent(by weight) of grain mixtures and meatmixtures, respectively (Enns, Goldman,& Cook, 1997). If vegetables accountfor a similar proportion of grain andmeat mixtures for children as for adults,then the observed higher intakes ofgrain mixtures would at least partiallyoffset the lower intakes of vegetables.Further research is needed to clarifythis issue. However, even when mixtureingredients are separated into theirrespective groups, at least 80 percent ofchildren had diets that did not meet thePyramid recommendations for servingsof vegetables (USDA, 2000b).

Despite Pyramid recommendations toeat both dark-green leafy vegetablesand legumes “several times a week,”children ate only one-tenth of a servingfrom either category on any given day.Children’s intakes of fried whitepotatoes were higher in 1994-96, 1998than in 1977-78, and the percentages ofchildren using tomatoes rose between1977-78 and 1994-96, 1998. Children

Table 3. Trends and changes in girls’ (6 to 11 years) percentages using items fromselected food groups

Percentage usingFood group 1977-78 1989-91 1994-96, 1998 Change1 Trend2

Grain products 99 100†3 99†

Yeast breads and rolls 79 73 71 -8Ready-to-eat cereals 51 48 46Cakes, cookies, pastries, pies 51 45 55Crackers, popcorn, pretzels, corn chips 18 22 37 +19Mixtures mainly grain 29 40 46 +17

Vegetables 87 81 82White potatoes 54 51 51 Fried white potatoes 30 37 39 +9Dark-green vegetables 7 7 5Deep-yellow vegetables 9 10 11Tomatoes 21 27 33 +12Green beans 14 8 8 -6Corn, green peas, lima beans 24 17 15 -9

Fruits 62 67 62Citrus juices 31 24 22 -9Apples 18 21 16Melons and berries 4 3 7Noncitrus juices and nectars 6 16 15 +9

Milk and milk products 95 93 90 -5Fluid milk 90 82 76 -14 * Whole milk 58 44 33 -25 ** Lowfat milk 17 39 38 +21 Skim milk 4 5 8 +5Milk desserts 24 21 22Cheese 17 28 32 +14

Meat, poultry, and fish 94 90 86 -7Beef 35 22 20 -15Pork 20 15 10 -10 *Frankfurters, sausages, luncheon meats 32 34 33Chicken 16 18 20Fish and shellfish 8 7 6Mixtures mainly meat, poultry, fish 31 36 33

Eggs 22 21 13 -8Legumes 13 15 11Fats and oils 54 56 49Sugars and sweets 56 55 60

Candy 11 16 29 +18 *Beverages 60 60 73 +12


1Change = percentages in 1977-78 and 1994-96, 1998 are significantly different at p < 0.001.2Trend = percentage rose or fell progressively from 1977-78 through 1989-91 to 1994-96, 1998.3Estimate is based on small sample size or coefficient of variation > 30 percent.†Value is between 99.5 and 100.* = trend significant at p < 0.05.** = trend significant at p < 0.01.


had both lower intakes and lowerpercentages using the subgroups “greenbeans” and “corn, green peas, limabeans” in 1994-96, 1998, comparedwith 1977-78.

Aside from the observed changes inintakes of noncitrus juices and nectars,few changes occurred in fruit consump-tion: Between 1977-78 and 1994-96,1998, the percentage using citrus juicesfell among girls. In 1994-96, 1998, only24 percent of girls and 23 percent ofboys consumed the number of servingsof fruit recommended in the FoodGuide Pyramid based on their caloricintake (USDA, 2000b).

Among milk and milk productssubgroups, children’s intakes of somehigh-fat items (e.g., whole milk)decreased and others (e.g., cheese)increased. Notably, milk intakes shiftedaway from whole milk.3 Decreasingtrends were seen in girls’ intakes ofwhole milk and in the percentages ofboth girls and boys using whole milk.Intakes of lower fat milks (2-percent,1-percent, and skim) surpassed thoseof whole milk in 1989-91 for boys andin 1994-96, 1998 for girls. Althoughthe percentages of children drinkingskim milk more than doubled between1977-78 and 1994-96, 1998, they stillremained low (8 to 9 percent), as didtheir intakes of skim milk (29 to 33 g,or about 1 fluid ounce). None of theshifts in intakes of lower fat milks orpercentages using them qualified as atrend.

3Another shift occurred that can be seen bysumming the milk subgroup intakes (whole,lowfat, and skim) in a given survey and dividingby the intake of total fluid milk. A greaterproportion of total fluid milk was allocated to aspecific fat level in later years than in 1977-78.The increase may indicate a greater awareness ofthe fat level of milk, since the ability to classifyfluid milk as whole, lowfat, or skim depends oninformation provided by respondents. Milkwhose fat level was not specified was includedunder total fluid milk but not in any of thesubgroups.

Table 4. Trends and changes in boys’ (6 to 11 years) percentages using items fromselected food groups

Percentage usingFood group 1977-78 1989-91 1994-96, 1998 Change1 Trend2

Grain products 100†3 100†3 993

Yeast breads and rolls 81 68 69 -12Ready-to-eat cereals 52 51 52Cakes, cookies, pastries, pies 52 39 52Crackers, popcorn, pretzels, corn chips 16 22 34 +18Mixtures mainly grain 26 46 45 +19

Vegetables 85 80 79White potatoes 56 46 49 Fried white potatoes 31 31 38Dark-green vegetables 5 5 6Deep-yellow vegetables 10 8 12Tomatoes 18 28 39 +21 **Green beans 13 10 7 -6Corn, green peas, lima beans 24 16 14 -10


Milk and milk products 94 90 92Fluid milk 90 79 79 -10 Whole milk 58 40 31 -27 * Lowfat milk 17 41 43 +26 Skim milk 3 5 9 +6Milk desserts 22 18 25Cheese 15 25 32 +17 **

Meat, poultry, and fish 95 88 88 -7Beef 33 18 22 -11Pork 22 15 12 -10Frankfurters, sausages, luncheon meats 33 30 36Chicken 17 18 20Fish and shellfish 7 9 5Mixtures mainly meat, poultry, fish 31 35 36

Eggs 23 20 16 -8Legumes 14 9 10Fats and oils 55 46 47Sugars and sweets 56 49 60

Candy 9 16 29 +20 **Beverages 62 64 74 +12

Tea 15 11 9Fruit drinks and ades 27 27 39 +12Carbonated soft drinks 31 38 47 +16

1Change = percentages in 1977-78 and 1994-96, 1998 are significantly different at p < 0.001.2Trend = percentage rose or fell progressively from 1977-78 through 1989-91 to 1994-96, 1998.3Estimate is based on small sample size or coefficient of variation > 30 percent.†Value is between 99.5 and 100.* = trend significant at p < 0.05.** = trend significant at p < 0.01.

2002 Vol. 14 No. 2 63

On the other hand, an increasing trendin the percentage using cheese was seenfor boys, while the increase for girlsdid not meet the definition of a trend.Cheese intakes were higher in 1994-96,1998 than in 1977-78 for both girlsand boys. Because cheese is a commoncomponent in both grain and meatmixtures, estimates for cheese wouldbe even higher if the cheese that wasan ingredient in these mixtures wasincluded here. In 1994-96, 1998, only29 percent of girls and 40 percent ofboys consumed the number of servingsof dairy products recommended in theFood Guide Pyramid based on their age(USDA, 2000b).

The percentages of children using foodsfrom the meat, poultry, and fish groupwere lower in 1994-96, 1998 than in1977-78. Both intakes and percentagesof individuals using beef and porkseparately (i.e., not as part of a mixture)fell. In all three surveys, intakes of“mixtures mainly meat, poultry, fish”—such as beef stew, hamburgers, chickenpot pie, and tuna salad—accounted forthe largest share of intakes of totalmeat, poultry, and fish. Intakes andpercentages of individuals eating eggswere lower in 1994-96, 1998 than in1977-78 for both boys and girls.

In 1994-96, 1998, only 12 percent ofgirls and 21 percent of boys consumedthe number of servings of meat andmeat alternates recommended in theFood Guide Pyramid based on theircaloric intake (USDA, 2000b). It isnoteworthy that cooked dry beans(other than soybeans) and peas, whichmay be tabulated under either thevegetable group or the meat group,were tabulated under the meat group forthat analysis; otherwise, the percentagesconsuming the recommended numberof servings from the meat group wouldhave been even lower.

Percentages using candy exhibitedupward trends for both girls and boys.Children’s candy intakes were higher in1994-96, 1998 than in 1977-78. Fats,oils, and sugars are common ingredientsin foods; thus, estimates of intakes andpercentages using fats, oils, and sugarswould be higher if the amounts thatwere ingredients in other foods wereincluded here.

In 1994-96, 1998, intakes of discretion-ary fat and added sugars4—items fromthe tip of the Pyramid—were muchhigher than recommended (USDA,2000b). At that time, discretionary fatintake accounted for about 25 percentof calories for girls and boys age 6to 11. In a diet that meets all otherPyramid recommendations, discretion-ary fat intake would be expected to becloser to 15 percent of calories (USDA,1996). In 1994-96, 1998, childrenage 6 to 11 consumed 21 to 23 tea-spoons of added sugars in a dietproviding around 1,800 to 2,000calories. The Pyramid suggests thatAmericans try to limit their addedsugars to 6 teaspoons a day if theyeat about 1,600 calories, 12 teaspoonsat 2,200 calories, or 18 teaspoons at2,800 calories (USDA, 1996).

Energy Out of BalanceOver roughly the same period coveredby the present analysis, the percentagesof 6- to 11-year-olds in the UnitedStates who were overweight5 nearlydoubled—from 7 to 8 percent in 1976-80 to 13 to 15 percent in 1988-94(U.S. Department of Health and HumanServices [DHHS], 2000). The increas-ing prevalence of overweight is of

4For definitions of discretionary fat and addedsugars, see appendix D in Pyramid Servingstable set 1 (USDA, 2000b).

5Overweight is defined as body mass index(BMI) at or above the sex- and age-specific 95thpercentile BMI cutoff points from the revisedCDC Growth Charts: United States (Kuczmarskiet al., 2000).

concern for many reasons, includingthe increasing incidence and prevalenceof Type II diabetes mellitus amongoverweight and obese children (Ameri-can Diabetes Association, 2000).

In the face of increasing overweight,one would expect to see either increas-ing energy intake, decreasing energyexpenditure, or both. In the presentanalysis, no significant trends orchanges were seen in energy intakesbetween 1977-78 and 1994-96, 1998(table 5). Energy intake stayed aboutthe same for girls and dipped in 1989-91 for boys.

Findings of underreporting in surveys,which are often but not always higheramong overweight respondents, mightlead one to speculate that the lack of atrend in energy intake could be due toincreased underreporting over time asa function of increased obesity. On theother hand, methodological improve-ments in the Agricultural ResearchService’s 24-hour recall have addressedseveral issues that are consideredimportant in obtaining complete intakedata (see “Design and Methods”).Using CSFII data, Krebs-Smith et al.(2000) identified low energy reportersby first estimating basal metabolic rate(BMR)6 based on self-reported bodyweight, sex, and age and then compar-ing the BMR estimates with a cutofflevel.7 They found that the percentageof adults who were low energy report-ers was lower in 1994-96 (15 percent)than in 1989-91 (25 percent). Theyalso found less underreporting amongchildren than among adults: Only 7.8percent of children age 2 to 11 in theCSFII 1994-96 were found to be low

6BMR was estimated by using the formuladeveloped by Schofield (1985).

7Eighty percent of BMR was the cutoff levelused. That level was proposed by Goldberg et al.(1991) as the lower limit of plausible energyintake for a single individual with 2 days ofintake data and 99.7 percent confidence limits.


energy reporters (S.M. Krebs-Smith,personal communication, March 8,2002). Livingstone and Robson (2000)have stated that determining whether achild’s energy intake is implausiblylow should take into account detailedinformation on the child’s activitylevel; however, such information isnot available from the three surveysin the present analysis.

Inactivity is probably a strong factor inthe increased prevalence of overweightin the United States (Weinsier, Hunter,Heini, Goran, & Sell, 1998). On anygiven day in 1994-96, 1998, 27 percentof girls and 32 percent of boys age 6to 11 watched 4 or more hours oftelevision or videos; 38 percent of girlsand 40 percent of boys watched 2 to3 hours; 35 percent of girls and 29percent of boys watched 1 hour orless (unpublished data). For children,accumulating at least 60 minutes ofmoderate physical activity on most orall days of the week has been identifiedas a goal to promote better long-termhealth (DHHS, 2001; USDA & DHHS,2000).

Energy-Providing Nutrients(Macronutrients)A trend toward higher carbohydrateintakes was evident among boys age 6to 11. For girls, carbohydrate intakewas 38 g per day higher in 1994-96,1998 than in 1977-78, although the p-value criterion for a trend was not met.For both girls and boys, protein and fatintakes were lower in 1994-96, 1998than in 1977-78.

These shifts in macronutrient intakesbetween 1977-78 and 1994-96, 1998were reflected in trends toward a lowerproportion of food-energy intake fromfat and a higher proportion fromcarbohydrate. Children’s percentage ofcalories from protein was also lower in1994-96, 1998 than in 1977-78, but thedecrease only reached trend status forboys. The proportion of energy from fat

in children’s diets in 1994-96, 1998(33 percent) was still higher than thatrecommended by the Dietary Guide-lines for Americans: 30 percent ofcalories or less (USDA & DHHS,2000). At 12 percent of calories,saturated fat intakes still exceeded therecommendation of less than 10 percentof calories.

Although the shifts in the proportionof energy intake from fat and carbohy-drate appear to have brought the macro-nutrient proportions in the average dietcloser to the recommended levels, acloser examination is less encouraging.The observed decrease in the percent-age of calories from fat is more dueto the increase in calories from carbo-hydrate than to the decrease in fatintake. Fat intake decreased by about100 kcal or less, but carbohydrateintake in-creased by about 150 to200 kcal, based on estimates in table 5multiplied by Merrill and Watt’s (1973)general conversion factors of 9 kcal/gfor fat and 4 kcal/g for carbohydrate.

Vitamins, Minerals, andOther Dietary ComponentsFor girls age 6 to 11, the only trendin vitamin or mineral intakes was adecrease in intake of vitamin B12;changes included higher intakes ofthiamin and iron in 1994-96, 1998 thanin 1977-78 (table 5). For boys, therewere increasing trends in intakes ofthiamin and iron; additional changesincluded higher intakes of vitamin C,riboflavin, niacin, and vitamin B6 anda lower intake of vitamin B12.

Mean dietary fiber intakes in 1994-96,1998 were 12 g for girls and 14 g forboys (unpublished data). One currentrecommendation suggests that fiberintakes for children should equal “ageplus 5 grams per day” (Williams,Bollella, & Wynder, 1995) (e.g., 13 gfiber for an 8-year-old). Observedincreases in carbohydrate intakeswere paralleled neither by significant

increases in dietary fiber intakes nor byincreases in overall intakes of fiber-richfoods.

Summary andRecommendations

A French proverb states, “The morethings change, the more they remainthe same.” The survey data used inpreparing this article span nearly aquarter of a century. Those who were6 to 11 years old in 1977-78 at the timeof the first survey used in this studywere 22 to 32 years old in 1994-96,1998—old enough to be the parentsof the children in the third survey usedhere. Although children’s food intakeshave changed in various ways overtime, the list of improvements thatare still needed has remained nearlyidentical.

Children’s diets exhibited trends towardlarge increases in intakes of soft drinksas well as decreases in intakes of totalfluid milk that were driven by decreasesin whole milk. Some other shifts wereto higher intakes of grain products(especially grain mixtures), crackers/popcorn/pretzels/corn chips, friedpotatoes, noncitrus juices/nectars,lowfat milk, skim milk, cheese, candy,and fruit drinks/ades. Other shifts wereto lower intakes of yeast breads/rolls,green beans, corn/green peas/limabeans, beef, pork, and eggs.

Despite those shifts in intakes, most ofthe take-home messages about how toimprove children’s diets remain thesame:• Eat more whole grains.• Eat more vegetables, especially

dark-green and deep-yellowvegetables.

• Eat more fruits—both citrus andnoncitrus, with an emphasis onwhole fruits rather than juices.

• Eat more legumes.

2002 Vol. 14 No. 2 65

Table 5. Trends and changes in girls’ and boys’ (6 to 11 years) mean intakes of foodenergy and selected nutrients and mean percentages of calories from protein, fat,and carbohydrate

IntakeFood group 1977-78 1989-91 1994-96, 1998 Change1 Trend2

Girls n=2,101 n=722 n=969

Energy (kcal) 1,806 1,832 1,825Protein (g) 69.3 67.2 62.7 -6.6Fat (g) 77.8 69.8 66.8 -11.0Carbohydrate (g) 211.9 241.6 250.0 +38.1Protein (% kcal) 15.5 15.1 13.9 -1.6Fat (% kcal) 38.2 33.8 32.6 -5.6 *Carbohydrate (% kcal) 47.4 52.6 54.9 +7.6 **Vitamin A (IU) 4,822 5,225 4,475Vitamin C (mg) 86 90 95Thiamin (mg) 1.32 1.53 1.48 +0.16Riboflavin (mg) 1.93 2.00 1.91Niacin (mg) 16.7 18.6 18.1Vitamin B6 (mg) 1.46 1.58 1.52Vitamin B12 (Fg) 5.36 4.34 3.87 -1.49 *Calcium (mg) 906 916 865Phosphorus (mg) 1,184 1,215 1,138Magnesium (mg) 230 230 219Iron (mg) 10.7 13.0 13.8 +3.1

Boysn=2,006 n=754 n=1,031

Energy (kcal) 1,950 1,891 2,050Protein (g) 75.6 70.1 71.2 -4.4Fat (g) 84.7 72.7 75.1 -9.6Carbohydrate (g) 226.2 245.5 279.6 +53.3 **Protein (% kcal) 15.7 15.0 14.0 -1.6 *Fat (% kcal) 38.5 34.2 32.6 -6.0 **Carbohydrate (% kcal) 46.8 52.0 54.8 +8.0 **Vitamin A (IU) 5,056 4,902 5,242Vitamin C (mg) 87 98 103 +16.0Thiamin (mg) 1.46 1.59 1.77 +0.31 *Riboflavin (mg) 2.11 2.11 2.28 +0.17Niacin (mg) 18.4 19.4 21.5 +3.1Vitamin B6 (mg) 1.59 1.62 1.84 +0.26Vitamin B12 (Fg) 5.88 4.42 4.53 -1.35Calcium (mg) 967 978 984Phosphorus (mg) 1,278 1,261 1,292Magnesium (mg) 244 233 249Iron (mg) 11.5 13.7 16.6 +5.1 **

1Change = mean intakes (or percentages) in 1977-78 and 1994-96, 1998 are significantly different atp < 0.001.2Trend = mean intake (or percentage) rose or fell progressively from 1977-78 through 1989-91 to 1994-96,1998.* = trend significant at p < 0.05.** = trend significant at p < 0.01.

• Shift to lean meats and meatalternates.

• Drink more skim or 1-percent milk,or eat more lowfat dairy products,or include plenty of nondairysources of calcium.

• Decrease the amount of fat used incooking.

The amount of discretionary fat andadded sugars in children’s diets is muchhigher than is recommended by theFood Guide Pyramid. Children’s dietswould benefit overall from loweringintakes of “empty-calorie” foods andbeverages that are high in fats andsugars but provide few other nutrients.In addition, when choosing amongmore nutrient-dense foods, childrenwould do well to shift toward itemslower in fat and sugar. Increases inintakes of foods high in fiber andcomplex carbohydrates—such aswhole grains, vegetables, fruits otherthan fruit juices, and legumes—couldlead to a diet lower in fat and addedsugars and higher in fiber and complexcarbohydrates. If such a change led toa lower overall energy intake, weightmaintenance or loss would be madeeasier. Because widespread inactivityhas been identified as a factor in thenational epidemic of overweight,increased activity should beencouraged.

Nutrition education can successfullychange dietary behavior amongelementary school-aged children, andfactors leading to the effectiveness ofnutrition education have been identified(Contento et al., 1995). Resources mustbe committed on every level—national,State, local, community, school, andfamily—to help children eat morehealthfully and become more active.


References

American Diabetes Association. (2000). Type 2 diabetes in children andadolescents. Diabetes Care, 23(3), 381-389.

Contento, I., Balch, G.I., Bronner, Y.L., Lytle, L.A., Maloney, S.K., Olson, C.M.,et al. (1995). The effectiveness of nutrition education and implications for nutritioneducation policy, programs, and research: A review of research. Journal ofNutrition Education, 27(6); special issue.

Enns, C.W., Goldman, J.D., & Cook, A. (1997). Trends in food and nutrientintakes by adults: NFCS 1977-78, CSFII 1989-91, and CSFII 1994-95. FamilyEconomics and Nutrition Review, 10(4), 2-15.

Federation of American Societies for Experimental Biology, Life SciencesResearch Office. (1995). Third Report on Nutrition Monitoring in the UnitedStates: Volume 1. Prepared for the Interagency Board for Nutrition Monitoring andRelated Research. Washington, DC: U.S. Government Printing Office. RetrievedMay 21, 2002, from http://www.cdc.gov/nchs/data/tronm.pdf.

Guenther, P.M. (1986). Beverages in the diets of American teenagers. Journal ofthe American Dietetic Association, 86(4), 493-499.

Goldberg, G.R., Black, A.E., Jebb, S.A., Cole, T.J., Murgatroyd, P.R., Coward,W.A., et al. (1991). Critical evaluation of energy intake data using fundamentalprinciples of energy physiology: 1. Derivation of cut-off limits to identify under-recording. European Journal of Clinical Nutrition 45, 569-581.

Harnack, L., Stang, J., & Story, M. (1999). Soft drink consumption among USchildren and adolescents: Nutritional consequences. Journal of the AmericanDietetic Association, 99(4), 436-441.

Krebs-Smith, S.M., Graubard, B.I., Kahle, L.L., Subar, A.F., Cleveland, L.E., &Ballard-Barbash, R. (2000). Low energy reporters vs others: A comparison ofreported food intakes. European Journal of Clinical Nutrition, 54(4), 281-287.

Kuczmarski, R.J., Ogden, C.L., Grummer-Strawn, L.M., Flegal, K.M., Guo, S.S.,Wei, R., et al. (2000). CDC Growth Charts: United States (Advance Data fromVital and Health Statistics, No. 314). Hyattsville, MD: National Center for HealthStatistics. Retrieved May 21, 2002, from http://www.cdc.gov/nchs/data/ad/ad314.pdf.

Livingstone, M.B.E., & Robson, P.J. (2000). Measurement of dietary intake inchildren. Proceedings of the Nutrition Society, 59(2), 279-293.

Ludwig, D.S., Peterson, K.E., & Gortmaker, S.L. (2001). Relation betweenconsumption of sugar-sweetened drinks and childhood obesity: A prospective,observational analysis. Lancet, 357, 505-508.

Merrill, A.L., & Watt, B.K. (1973). Energy value of foods—basis and derivation.U.S. Department of Agriculture, Agriculture Handbook No. 74, sl. rev.

2002 Vol. 14 No. 2 67

SAS® (Version 8.2) [computer software]. (1999). Cary, NC: SAS Institute.

Schofield, W.N. (1985). Predicting basal metabolic rate, new standards and reviewof previous work. Human Nutrition Clinical Nutrition, 39C(Suppl. 1), 5-41.

Shah, B.V., Barnwell, B.G., & Bieler, G.S. (1997). SUDAAN (Version 7.5.1)[computer program]. Research Triangle Park, NC: Research Triangle Institute.

Tippett, K.S., & Cypel, Y.S. (Eds.). (1998). Design and Operation: TheContinuing Survey of Food Intakes by Individuals and the Diet and HealthKnowledge Survey 1994-96. U.S. Department of Agriculture, AgriculturalResearch Service, Nationwide Food Surveys Rep. No. 96-1; NTIS No. PB 98-137268. Retrieved May 21, 2002, from http://www.barc.usda.gov/bhnrc/foodsurvey/Dor.html.

Tippett, K.S., Enns, C.W., & Moshfegh, A.J. (2000). Food consumption surveys inthe U.S. Department of Agriculture. In F.J. Francis (Ed.), Encyclopedia of FoodScience and Technology (2nd. ed., pp. 889-897). New York: Wiley.

Tippett, K.S., Mickle, S.J., Goldman, J.D., Sykes, K.E., Cook, D.A., Sebastian,R.S., et al. (1995). Food and Nutrient Intakes by Individuals in the United States,1 day, 1989-91. U.S. Department of Agriculture, Agricultural Research Service,Continuing Survey of Food Intakes by Individuals 1989-91, Nationwide FoodSurveys Rep. No. 91-2; NTIS No. PB 95-272746.

U.S. Department of Agriculture. (1983). Food Intakes: Individuals in 48 States,Year 1977-78. U.S. Department of Agriculture, Human Nutrition InformationService, Nationwide Food Consumption Survey 1977-78, Rep. I-1; NTIS No.PB91-103523.

U.S. Department of Agriculture. (1987). CSFII: Women 19-50 Years and TheirChildren 1-5 Years, 4 Days, 1985. U.S. Department of Agriculture, HumanNutrition Information Service, Nationwide Food Consumption Survey, ContinuingSurvey of Food Intakes by Individuals, Rep. 85-4; NTIS No. PB88-110101.

U.S. Department of Agriculture. (1993). Food and Nutrient Intakes by Individualsin the United States, 1 Day, 1987-88. U.S. Department of Agriculture, HumanNutrition Information Service, Nationwide Food Consumption Survey, Rep. 87-I-1; NTIS No. PB 94-168325.

U.S. Department of Agriculture, Agricultural Research Service. (1999). Food andNutrient Intakes by Children 1994-96, 1998; ARS Food Surveys Research GroupTable Set 17. Retrieved May 21, 2002, from http://www.barc.usda.gov/bhnrc/foodsurvey/pdf/scs_all.pdf.

U.S. Department of Agriculture, Agricultural Research Service. (2000a).Continuing Survey of Food Intakes by Individuals 1994-96, 1998 [CD-ROM].NTIS No. PB2000-500027.


U.S. Department of Agriculture, Agricultural Research Service. (2000b). PyramidServings Intakes by U.S. Children and Adults: 1994-96, 1998; ARS CommunityNutrition Research Group Table Set No. 1. Retrieved May 21, 2002, from http://www.barc.usda.gov/bhnrc/cnrg/tables.pdf.

U.S. Department of Agriculture, Center for Nutrition Policy and Promotion.(1996). The Food Guide Pyramid (sl. rev., Home and Garden Bulletin No. 252).Washington, DC: U.S. Government Printing Office. Retrieved May 21, 2002, fromhttp://www.usda.gov/cnpp/Pubs/Pyramid/fdgdpyr1.pdf.

U.S. Department of Agriculture and U.S. Department of Health and HumanServices. (2000). Nutrition and Your Health: Dietary Guidelines for Americans(5th ed., Home and Garden Bulletin No. 232). Washington, DC: U.S. GovernmentPrinting Office. Retrieved May 21, 2002, from http://www.usda.gov/cnpp/DietGd.pdf.

U.S. Department of Health and Human Services. (2001). The Surgeon General’sCall to Action to Prevent and Decrease Overweight and Obesity. Rockville, MD:Public Health Service, Office of the Surgeon General. Retrieved May 21, 2002,from http://www.surgeongeneral.gov/topics/obesity/calltoaction/CalltoAction.pdf.

U.S. Department of Health and Human Services, National Center for HealthStatistics. (2000). Table 69: Overweight children and adolescents 6-17 years ofage, according to sex, age, race, and Hispanic origin: United States, selected years1963-65 through 1988-94. In Health United States 2000 with AdolescentChartbook . Retrieved May 21, 2002, from http://www.cdc.gov/nchs/data/hus/hus00.pdf.

Weinsier, R.L., Hunter, G.R., Heini, A.F., Goran, M.I., & Sell, S.M. (1998). Theetiology of obesity: Relative contribution of metabolic factors, diet, and physicalactivity. American Journal of Medicine 105(2), 145-150.

Williams, C.L., Bollella, M., & Wynder, E.L. (1995). A new recommendation fordietary fiber in childhood. Pediatrics, 96(5), 985-988.

Wyshak, G. (2000). Teenaged girls, carbonated beverage consumption, and bonefractures. Archives of Pediatric and Adolescent Medicine, 154, 610-613.

2002 Vol. 14 No. 2 69

Trends in Children’s Consumptionof Beverages: 1987 to 1998

Beverages contribute to the overall quality of children’s dietary intake. This studyexamined trends in beverage consumption for children age 1 to 19 years for three2-year periods between 1987 and 1998. Intakes were collected from a nationallyrepresentative sample of households. A 2-week diary by the National FamilyOpinion Research/Beverage Unit’s Share of Intake Panel during 1987-88, 1992-93, and 1997-98 was used to collect these intakes. During these periods, theconsumption of carbonated soft drinks decreased significantly for children age1 to 5 years—83.5, 78.2, and 72.1 percent, respectively, whereas consumptionof milk remained stable for all groups, and consumption of lowfat milk becamemore prevalent than consumption of whole milk. The quantity of milk consumedincreased for children age 1 to 5 years (11.6 to 13.5 oz/day), while the quantityof carbonated soft drinks consumed decreased (5.2 to 3.7 oz/day). The quantityof fruit drinks increased for all age groups. This study is useful to those who aredeveloping strategies to improve the overall quality of children’s diets. Furtherresearch is needed to investigate the effect children’s beverage consumption hason their health and to evaluate the effect of beverage consumption on total dietaryintake.

Yi Kyung Park, MSMichigan State University

Emily R. Meier, BSMichigan State University

Peri Bianchi, PhDSpectrum Health

Won O. Song, PhD, MPH, RDMichigan State University

hildren consume fewer meals atthe dining table and more mealsand snacks away from home.

Only 36 percent of respondents to arecent study on eating habits reportedthat they ate together as a family fiveor more nights a week (Anonymous,2000). For teenagers, 5 percent of out-of-home eating occasions consisted ofa trip to a convenience store—mostoften for a beverage purchased from avending machine (Anonymous, 1999b).Throughout the day, children consumea variety of beverages, which ofteninclude milk, juice, fruit drinks, carbo-nated soft drinks, powdered soft drinks,and water.

To implement strategies to improvethe health of children, the nutritioncommunity needs to understand theeffect of beverage consumption onnutrition, in terms of the frequency andquantity of each beverage consumed,as well as the relationship of beverages

to food consumption and health.However, it is somewhat difficult todetermine the contribution of eachbeverage to the overall diet, especiallychildren’s diets.

Efforts to promote milk consumptionhave increased, and their effects onhealth and nutrition have yet to beunderstood fully. Calcium intake islow for adolescent girls, a particularconcern because calcium absorptionis at its highest during adolescence(Amschler, 1999). When calcium intakeis inadequate early in life, the risk ofosteoporosis increases (NationalInstitutes of Health [NIH], 1994). Somestudies have reported that consumptionof carbonated soft drinks has increased;this may have a significant effect onnutrition (Borrud, Enns, & Mickel,1997).

Research suggests that caffeine, whichis more frequently found in cola

C


beverages than in other soft drinks,may reduce calcium retention (Weaver,Proulx, & Heaney, 1999). Also, thephosphoric acid in carbonatedbeverages (cola and others) interfereswith the metabolism of calcium (Calvo,1994; Wyshak, 2000; Wyshak &Frisch, 1994). Other researchers,however, disagreed with these findings(The Pediatric Forum, 2001). Heaneyand Rafferty (2001) concluded thatthe net effect of caffeine on calciumeconomy was negligible.

Finally, no conclusive evidence hasindicated that carbonated soft drinksare the “cause” of low intakes of otherbeverages or that they are displacingother beverages in the diet, althoughsome studies indicate a very suggestiverelationship between higher quantitiesof carbonated soft drinks and lowerquantities of milk consumption(Bowman, 1999; Harnack, Stang, &Story, 1999).

The debate over the consumption offruit juice remains, having been arguedto have effects on children’s health thatare both favorable (Doucette & Dwyer,2000; Skinner & Carruth, 2001) andunfavorable (Dennison, Rockwell, &Baker, 1997; Tanasescu, Ferris,Himmelgreen, Rodriquez, & Perez-Escamilla, 2000). Regardless, 100-percent fruit juice is an excellent sourceof many essential vitamins and mineralsand is consumed by a significantnumber of U.S. children (Ballew,Kuester, & Gillespie, 2000).

In recent years, fruit drinks—oftenmade with only 5 to 10 percent fruitjuice—have emerged as a growingcomponent of the American diet. In1997, sales of fruit drinks surpassedsales of 100-percent fruit juices(Sfiligoj, 1998). Fruit drinks, despitetheir higher sugar content, comparedwith fruit juices, provide about 17percent of vitamin C for children age 2to 5 and are the second greatest source

of vitamin C for children of all ages(Subar, Krebs-Smith, Cook, & Kahle,1998). Although fruit drinks couldcontribute to improved vitamin intakes,their effect on nutrition is not under-stood fully.

Beverages are a significant portion ofthe American diet, contributing to theintake of various nutrients as well asadded sugars. The Dietary Guidelinesfor Americans recommend thatAmericans “Choose beverages andfoods to moderate their intake ofsugars” (U.S. Department of Agri-culture [USDA] and Department ofHealth and Human Services [DHHS],2000). However, a recent study foundthat among children age 2 to 17, 38 to56 percent of the added sugars theyconsumed came from beverages(Guthrie & Morton, 2000). Becausechildhood obesity is rising, moreinformation is needed to investigatethe changes in children’s diets overtime. The purpose of this study wasto investigate trends in children’sbeverage consumption (both prevalenceand quantity) over a 10-year period.

Methods

Survey DesignData were obtained from the NationalFamily Opinion Research/BeverageUnit’s Share of Intake Panel (SIP)(National Family Opinion World Group[NFO], 1999; National Soft DrinkAssociation [NSDA], 1999), a syndi-cated marketing research program atthe national level. Since 1980, SIP hasused mail surveys, which are primarilypurchased by industry members, tomonitor beverage consumption. Thesurvey instrument is a 2-week diarythat collects all beverage intakes(excluding tap water) for all membersof the selected household. Participantsare told to exclude tap water, but toinclude bottled water. Tap water isdifficult for consumers to quantify,

even though its contribution tobeverage consumption may besignificant. In this study, milk intakereflects only beverages; milk consumedwith cereal was not included, nor wasmilk in other forms of dairy products(e.g., cheese).

SIP samples 3,000 individualsquarterly, for an annual total of 12,000individuals. Mailings to 4,498 house-holds in each quarter are staggeredweekly to ensure coverage across thewhole year. The sampling is done at thehousehold level and balanced at theindividual level by weighting the dataquarterly to be representative of theU.S. population (age within gender,household income and size, region,and market size).

In addition to demographic information,the SIP diary collects the followinginformation for each beverage con-sumed: brand and flavor, beverageattributes (e.g., diet, regular, sugar-free), time of day and month whenconsumed, type and size of container inwhich it was purchased, class of tradewhere drink was purchased, type ofcontainer from which it was consumed,where it was consumed (home, away,etc.), temperature of beverage whenconsumed (hot, cold), and quantity (inounces) consumed at one occasion.

Because a mail survey was used tocollect the data, all initial instructionswere given in writing. Households wereinstructed how to quantify beverages(ounces or cups) by providing themwith two-dimensional food models. Forthe children who were unable to keeptheir own records, another familymember did so.

Data AnalysisWe analyzed data from children whowere age 1 to 19 years in 1987-88(n=4,143), 1992-93 (n=2,748), and1997-98 (n=2,397). Beverages thatmade the greatest contribution to total

2002 Vol. 14 No. 2 71

beverage intake were used in thisinvestigation: milk (whole, lowfat,and skim/buttermilk), carbonated softdrinks (regular and diet), fruit juices,fruit drinks, powdered soft drinks(unsweetened, presweetened, sugar-sweetened, and sugar-free), and tea(hot, cold, herbal, and ready-to-drink).“Lowfat” milk consisted of any milknot specified as either whole milk orskim milk (i.e., 1% and 2% milks).The classification of fruit juice wasbased on the respondents’ perceptionsand knowledge about the variety ofbeverages. “Other” beveragesconsisted of coffee, breakfast drinks,beer, and other forms of alcohol. Fewerthan 5 percent of the children werereported to have consumed these otherbeverages; thus, the contribution ofthese beverages to the total volumeof beverage intake was minor andnot included in this analysis.

Of the children who drank thebeverages studied, the total quantitiesover each 2-week survey period wereconverted so that we could estimatemean daily consumption (in ounces/day) for each specific age/gendergroup. Chi-square tests for trendswere performed to observe changes inprevalence of beverage consumptionover the three survey periods;Bonferoni t-tests were performed totest for statistical significance in meandaily intake among survey periods.

Results

PrevalenceMilk, carbonated soft drinks, andjuices, respectively, were the mostcommonly consumed beverages acrossage groups (table 1). Seventy-six to 95percent of the children consumed milk;72 to 97 percent, carbonated softdrinks; and 53 to 79 percent, juices.The percentage of children reportingmilk consumption, however, remainedstatistically similar over the decade.

Table 1. Children drinking selected beverages, by selected years

1987-88 1992-93 1997-98Gender/age groups (n=4,143) (n=2,748) (n=2,397)

PercentMilk1

Children, 1-5 years 93.8 95.1 90.7Children, 6-9 years 95.2 94.8 94.5Boys, 10-14 years 92.2 92.2 87.0Girls, 10-14 years 91.7 88.9 83.4Boys, 15-19 years 87.2 81.7 80.9Girls, 15-19 years 75.9 75.5 78.1

Carbonated soft drinksChildren, 1-5 years* 83.5 78.2 72.1Children, 6-9 years 92.4 91.1 86.7Boys, 10-14 years 92.6 93.6 91.4Girls, 10-14 years 94.5 91.7 90.3Boys, 15-19 years 94.6 95.9 95.2Girls, 15-19 years** 96.8 94.8 81.9

JuicesChildren, 1-5 years 77.3 76.7 78.5Children, 6-9 years 74.1 68.4 68.9Boys, 10-14 years 66.6 61.4 54.2Girls, 10-14 years 69.5 67.4 63.0Boys, 15-19 years 65.8 66.9 56.8Girls, 15-19 years 63.5 63.9 53.4

Fruit drinksChildren, 1-5 years 53.3 55.4 61.0Children, 6-9 years 55.6 58.9 62.1Boys, 10-14 years 47.6 49.5 60.1Girls, 10-14 years 49.3 54.3 56.6Boys, 15-19 years 44.4 37.8 46.1Girls, 15-19 years 35.9 47.4 50.2

Powdered soft drinksChildren, 1-5 years 54.4 50.7 45.2Children, 6-9 years* 59.1 52.8 44.4Boys, 10-14 years 48.5 43.8 40.8Girls, 10-14 years 52.7 40.7 43.0Boys, 15-19 years 35.7 35.9 26.8Girls, 15-19 years 27.5 30.2 23.1

TeaChildren, 1-5 years** 32.6 21.2 17.3Children, 6-9 years* 37.3 23.4 20.3Boys, 10-14 years 42.0 35.6 37.1Girls, 10-14 years* 42.3 40.4 26.4Boys, 15-19 years 41.0 39.3 41.5Girls, 15-19 years 50.9 36.0 31.8

1Contains milk beverages only. Milk consumed with cereal or milk in other forms (e.g., cheese) is not included.* p<0.05, ** p<0.001 = significant difference in percentages across selected years.Note: Beverage consumption was gauged if a child consumed at least one serving of the beverage over a2-week period.


The prevalence of milk consumptionwas lowest for girls age 15 to 19,compared with the other age groups. Alittle more than three-fourths of 15- to19-year-old girls reported any milkconsumption during the three 2-weekreporting periods (76 percent in 1987-88, 76 percent in 1992-93, and 78percent in 1997-98).

From 1987-88 to 1997-98, while asignificant decline occurred in thepercentage of children in all age groupswho drank whole milk, a rise occurredin the percentage consuming lowfat andskim/buttermilk (fig. 1). By 1992-93,the percentage of children drinkinglowfat milk exceeded the percentagesof children drinking whole milk in allage groups. By 1997-98, 52 to 68percent of the children in the variousage/gender groups drank lowfat milk;

whereas, 22 to 47 percent of thechildren consumed whole milk. Thepercentage of children drinking skim/buttermilk increased over the decade,particularly among younger childrenand 15- to 19-year-old girls. About20 percent of girls age 10 to 19 yearsconsumed skim/buttermilk in 1997-98.

The percentage of children drinkingcarbonated soft drinks did not changefor most of the children age 6 to 9, girlsage 10 to 14, or boys age 10 to 14 and15 to 19 (table 1). There was, however,a significant decline in the percentageof children age 1 to 5 and girls age 15to 19 who consumed carbonated softdrinks. During 1987-88, 84 percent of1- to 5-year-olds consumed carbonatedsoft drinks; by 1997-98, 72 percentconsumed this type of beverage. During

1987-89, 97 percent of girls age 15 to19 consumed carbonated soft drinks,compared with 82 percent by 1997-98,a drop of 15 percentage points.

Over the survey period, the percentageof children who drank either regular ordiet carbonated soft drinks remainedstable, except for 1- to 5- and 15- to 19-year-old girls (fig. 2). The percentageof the youngest children (age 1 to 5years) who consumed regular carbo-nated soft drinks decreased signifi-cantly: dropping from 79 to 69 percent.The percentage of the older girls (15 to19 years) who drank regular or dietcarbonated soft drinks also decreasedsignificantly: from 90 to 77 percent in1987-88 (regular) and 35 to 18 percent,in 1997-98 (diet).

Children

, 1-5 y

ears*

*

Childre

n, 6-9 y

ears**

Boys,

10-14

years

**

Girls, 1

0-14 y

ears*

*

Boys,

15-19

years

*

Girls, 15

-19 ye

ars**

Children

, 1-5 y

ears

Children

, 6-9

years

*

Boys,

10-14

years

*

Girls, 10

-14 ye

ars

Boys,

15-19

years

Girls, 15

-19 ye

ars

Children

, 1-5 y

ears*

*

Children

, 6-9 y

ears**

Boys,

10-14

years

Girls, 10

-14 ye

ars

Boys,

15-19

years

Girls, 1

5-19 y

ears*

0

10

20

30

40

50

60

70

80

Per

cent

con

sum

ing

1987-88 1992-93 1997-98

Figure 1. Children consuming whole, lowfat, and skim milk/buttermilk,1 by selected years

Whole milk Lowfat milk2 Skim milk/buttermilk

1Contains milk beverages only. Milk consumed with cereal or milk in other forms (e.g., cheese) is not included.21% and 2% milks.*p<0.05, **p<0.001 = significant differences in percentages across the selected years.

2002 Vol. 14 No. 2 73

Over the survey periods, the percentageof children reporting juice consumptiondecreased slightly (table 1). However,the percentage of children among allage groups consuming fruit drinksincreased over the 10-year period—although none of the differences werestatistically significant. The number ofchildren consuming powdered softdrinks decreased significantly amongyounger children (6 to 9 years),dropping from 59 to 44 percent. Thenumber of children consuming tea alsodecreased significantly across thesurvey period for three age groups—children age 1 to 5 (33 to 17 percent),children age 6 to 9 (37 to 20 percent),and girls age 10 to 14 (42 to 26percent).

QuantityBoys consumed more milk than didgirls in the same age group, and youngchildren (1 to 9 years) drank more milkthan any other beverages across allthree periods. Mean daily milk intake

increased significantly for children age1 to 5: from 11.6 to 13.5 oz between1987-88 and 1997-98 (table 2).

Children age 1 to 5 years decreasedsignificantly their mean daily intake ofcarbonated soft drinks: from 5.2 to 3.7oz/day between 1987-88 and 1997-98.The amount consumed by the otherage/gender groups did not changesignificantly. Younger children (1 to 9years) drank more milk (12 to 14 oz/day) than carbonated soft drinks (4 to 8oz/day), but older children (10 to 19years) consumed more carbonated softdrinks (12 to 23 oz/day) than milk (9 to15 oz/day).

For children age 1 to 5 years, thequantity of fruit juice consumedincreased significantly, from 5.0 to7.0 oz/day during the 10-year period.For boys age 10 to 14 years, however,the quantity consumed decreasedsignificantly from 4.5 to 3.3 oz/day.The average amount of fruit juice

Figure 2. Children consuming carbonated soft drinks, by selected years

Children

, 1-5 y

ears*

Children

, 6-9

years

Boys,

10-14

years

Girls, 1

0-14 y

ears

Boys,

15-19

years

Girls, 15

-19 ye

ars*

Children

, 1-5 y

ears

Children

, 6-9 y

ears

Boys,

10-14

years

Girls, 1

0-14 y

ears

Boys,

15-19

years

Girls, 15

-19 ye

ars*

010

20

3040

50

60

7080

90

100

Per

cent

age

cons

umin

g

1987-88 1992-93 1997-98

*p<0.05 = significant difference in percentages across the selected years.

Regular carbonated soft drinks Diet carbonated soft drinks

. . . we found no significantdecline in the prevalence ofchildren’s milk consumption overthe past decade, nor did we findan increase in the prevalenceof carbonated soft drinkconsumption.


remained stable for the other age/gender groups, with no groupconsuming more than 7.0 oz/day,on average.

The mean daily intake of fruit drinksfor all age/gender groups increasedacross the decade. The quantitiesconsumed were similar for all agegroups and were slightly less than themean quantities of fruit juice consumed.The quantity of tea consumed wasnegligible (despite its prevalenceamong older children) and wastherefore not included in calculatingestimates of mean daily consumption.

While the mean intake of fruit drinksincreased over the decade, the averageamount of noncarbonated soft drinksmade from a powdered mix declinedsignificantly. This decrease was mostapparent for 15- to 19-year-old boys(from 4.6 to 2.1 oz/day).

Discussion

PrevalenceIn light of the conclusion made byHarnack and colleagues (1999) thatcarbonated beverages have displacedmilk consumption, we found nosignificant decline in the prevalenceof children’s milk consumption overthe past decade, nor did we find anincrease in the prevalence of carbon-ated soft drink consumption. An equallyimportant trend is that children tendedto shift from drinking whole milk tolower fat varieties over the decade,with lowfat milk being the favoritetype of milk to drink by 1997-98.

Carbonated beverages continue tobe a popular drink for children andadolescents. In 1997-98 carbonated softdrink intake was reported by 72 percentof even the youngest children (age 1 to

Table 2. Children’s mean daily intake1 of selected beverages,2 by selected years

1987-88 1992-93 1997-98 (n=4,143) ( n=2,748) (n=2,397)

(Ounces/day) Milk3

Children, 1-5 years 11.6a 12.5b 13.5b

Children, 6-9 years 12.2a 13.0b 12.1ab

Boys, 10-14 years 13.6 13.7 12.5Girls, 10-14 years 11.9 11.6 11.2Boys, 15-19 years 14.7 12.7 11.7Girls, 15-19 years 9.7 8.9 10.7

Carbonated soft drinksChildren, 1-5 years 5.2a 4.4b 3.7c

Children, 6-9 years 7.4 7.8 7.7Boys, 10-14 years 13.6 15.6 14.6Girls, 10-14 years 12.3 12.1 12.5Boys, 15-19 years 22.0 21.4 22.9Girls, 15-19 years 19.8 17.3 16.1

JuicesChildren, 1-5 years 5.0a 6.3b 7.0b

Children, 6-9 years 4.0 4.0 4.0Boys, 10-14 years 4.5a 3.5b 3.3b

Girls, 10-14 years 3.7 3.5 3.3Boys, 15-19 years 4.3 4.4 4.3Girls, 15-19 years 3.7 4.0 3.3

Fruit drinksChildren, 1-5 years 2.1a 2.8b 2.9b

Children, 6-9 years 2.2a 2.8b 2.9b

Boys, 10-14 years 3.1a 2.4b 3.7a

Girls, 10-14 years 2.3a 2.5a 3.6b

Boys, 15-19 years 2.9ab 2.4a 3.8b

Girls, 15-19 years 1.7a 3.5b 3.1b

Powdered soft drinksChildren, 1-5 years 4.0a 3.5b 3.5b

Children, 6-9 years 4.4a 4.8a 3.3b

Boys, 10-14 years 4.8a 3.4b 3.4ab

Girls, 10-14 years 4.1a 2.9b 3.4ab

Boys, 15-19 years 4.6a 4.1a 2.1b

Girls, 15-19 years 1.6a 2.3b 1.7ab

1Mean daily intakes are among those children who reported consumption.2Significance is noted only when quantities are significantly different between at least two of the three periodsfor an age/gender group. Groups that share the same letter, however, are not significantly different at p<0.05.3Contains milk beverages only. Milk consumed with cereal or milk in other forms (e.g., cheese) is not included.

2002 Vol. 14 No. 2 75

5 years), and it was considerablymore prevalent among preteens andadolescents during the 2-week reportingperiods in this study. These results aresomewhat higher than those previouslyreported in the Continuing Survey ofFood Intakes by Individuals (CSFII)(Federation of American Societies forExperimental Biology [FASEB], 1995;USDA, 1997). In the CSFII 1994-96,37 percent of 3- to 5-year-olds, 48 and44 percent of 6- to 11-year-old boysand girls, respectively, and 69 and 62percent of 12- to 19-year-old boys andgirls consumed carbonated soft drinksover a 2-day reference period. It ispossible that differences in study designbetween SIP and the CSFII may haveinfluenced the results. (CSFII used2 days of dietary recalls to indicateprevalence, compared with 2 weeksused by SIP.) The data in this studyshow no increase in the prevalenceof carbonated soft drinks during thereporting periods and show that thepercentage of children age 1 to 5and girls age 15 to 19 who reporteddrinking soft drinks actually decreasedover time.

QuantityThe mean daily intake of milk byadolescent girls has remained relativelystable—although well below recom-mended quantities. In 1997-98 theaverage daily milk intake by girls age10 to 14 and 15 to 19 years accountsfor about 30 percent of the DietaryReference Intake (DRI) for calcium(Baker et al., 2000; National Academyof Sciences [NAS], 1999). Ballew,Kuester, and Gillespie (2000) reporteda concern for girls with a strongpositive correlation between milkconsumption and adequate intakesof vitamins A and B12, folate, calcium,and magnesium. On the other hand, theamount of milk consumed by children1 to 5 years old increased significantlyand represents nearly 85 percent ofthe minimum recommended amount(2 to 3 8-oz servings of milk or its

equivalent) from the Food GuidePyramid dairy group with fluid milkalone (not including milk on cerealor in other forms, such as cheeses),indicating that the increase in meandaily intake is an encouraging change.

Among those children who reportedconsuming carbonated soft drinks, themean daily intake did not change formost age and gender groups. Otherstudies observed significant increasesin the quantity of carbonated soft drinksconsumed (FASEB, 1995; USDA,1997; Troiano, Briefel, Carroll, &Bialostosky, 2000). Harnack, Stang,and Story (1999) found a strongassociation between increasedconsumption of soft drinks anddecreased consumption of milk andjuice. They also reported that childrenwho drank 9 or more ounces ofcarbonated soft drinks per dayconsumed significantly more energythan those who drank less than 9 oz.Intake of carbonated beverages,especially colas, was implicated as arisk factor for bone fracture in somestudies (Goulding et al., 1998; Petridou,Karpathios, Dessypris, Simou, &Trichopoulos, 1997), although otherresearch does not suggest such a link(Heaney & Rafferty, 2001).

The trends we observed in teenagegirls’ consumption of milk may bea concern regarding the developmentand maintenance of healthy bones. Theimportance of milk/dairy consumptionshould be emphasized to encourageadequate calcium intakes. Girlscontinued to consume nearly twicethe amount of carbonated soft drinks,compared with milk; however, with thequantity of carbonated soft drink intakestable, and a decreased prevalence, nodata in our study support the theory thatcarbonated soft drinks are displacingmilk in children’s diets.

Some have suggested that carbonatedsoft drinks displace more healthful

. . . children tended to shift fromdrinking whole milk to lower fatvarieties over the decade, withlowfat milk being the favoritetype of milk to drink by 1997-98.


beverages (Tanasescu, Ferris,Himmelgreen, Rodriquez, & Perez-Escamilla, 2000; Cavadini, Siega-Riz,& Popkin, 2000). Despite the lack ofdata showing an increase in the quantityof carbonated soft drinks consumedover the decade, the average dailyintake of fruit juice remained fairlystable across most age groups,increased significantly by 2 oz/dayamong children age 1 to 5 years, anddecreased significantly by 1 oz/dayfor boys age 10 to 14 years. Based onFood Guide Pyramid recommendations(USDA & DHHS, 1995), these dataindicate that fruit juice accountedfor over half of the recommendedminimum number of servings per day(i.e., 2 6-oz servings of juice).

The quantity of fruit drinks consumedincreased significantly over the decade.Fruit drinks have been heavilymarketed to children and packagedfor handy snacking (Russo, 1998).Although they generally have higheradded sugar content than do fruit juices,many have been fortified with essentialvitamins and minerals. When onemanufacturer added calcium to fruitjuice and fruit drinks in 1998, thesales response was so strong thatother companies expanded calciumfortification of their product lines(Anonymous, 1999a). These beveragesmay make increasingly importantcontributions to children’s diets, andan increase in their consumption maybe beneficial if fortified fruit drinkswith reduced amounts of added sugarsare selected. Despite the benefits offruit drinks, the trend toward increasedconsumption and prevalence may bea concern if consumption of otherbeverages is decreased as a result.

Strengths/LimitationsWhile our data provide an overviewof changes in children’s consumptionpatterns over the past decade, strengthsas well as limitations of this studyshould be considered. A major strength

of our data is that 2 weeks ofinformation on consumed beverageswere collected in a journal instead ofon a recall basis. Respondents onlyrecorded their beverage intake and notfoods, as is required for a full foodrecord, so the respondents’ burden mayhave been reduced. Another strength isthat the methods of data collection forthis study have not changed in the pasttwo decades, making comparisons ofsurvey years straightforward. In someother surveys, collection methods havechanged for each survey period,complicating the results andinterpretation.

Although the samples were collected inthe same manner throughout the studyperiod—with specific attention given toobtaining a nationally representativesample—some sample response biasthat is inherent to all surveys may haveoccurred over time (Bingham & Day,1997). No oral instructions or three-dimensional food models were given toassist the respondents in quantifyingand classifying the beverages; thus, anyincongruence between the respondents’knowledge and factual definitionsremained.

Implications

In this study, it was not possible todetermine what the trend in totalbeverage intake is, because tap waterwas not included in the data collection,and it was not the study’s purpose. Itis simply not viable to say that onebeverage is displacing another, since itcould be merely increasing in additionto other fluids (e.g., tap water). Therole that each beverage type plays inoverall fluid consumption needs to beaddressed in future research endeavors.Theoretically, the method for datacollection is better in a randomlysampled population; however, theresults from other panel samplesyielded results similar to those in the

randomly sampled CSFII population.Although more research is needed toexamine specific populations at risk,children (and their parents if thechildren are too young to decidethemselves) have been somewhatsuccessful in making more healthfulfood selections (e.g., skim milk vs.whole milk). This verification ofsuccessful changes in dietary behaviorsencourages future nutrition educationefforts to include beverage selectionsas a part of dietary habits that promotehealthful lifestyles.

Children’s intake of calcium—estimated by the average amount ofmilk consumed—still falls below therecommended calcium intakes for allage groups. This study demonstratedsome discrepancy between recom-mendations for children’s dietary intakeand actual intake of certain beverages.Significant changes in children’s dietsover the 10-year period include anincrease in the amounts of milk andjuice consumed by younger children, adecrease in the quantity of carbonatedsoft drinks consumed by 1- to 5-year-old children and 15- to 19-year-oldgirls, as well as an increase in thequantity of fruit drinks consumedby children of all ages. Hence, it isimperative to assess the contributionthat beverages make to micro- andmacronutrient intakes.

The trends in beverage consumptionare only one area of overall dietaryintake that we explored. The effectsthat beverages have on nutrition andhealthful dietary patterns need tobe investigated further. Changes inbeverage consumption may be anindicator of overall changes in diet,but this has yet to be determined.Little is known about the lifestyles andhabits related to beverage consumption,which could play an important rolewithin trends. The trends in children’sbeverage consumption in the past 10 to15 years provide insight into potential

2002 Vol. 14 No. 2 77

educational programs that may helpimprove children’s dietary habits andhealth. Once detailed information isobtained regarding precise populationsat risk, educational strategies can beimplemented to encourage and facilitatechange.

Acknowledgments

The Food and Nutrition DatabaseResearch Center at Michigan StateUniversity and the National Soft DrinkAssociation provided funding for thisproject.

References

Amschler, D.H. (1999). Calcium intake in adolescents: An issue revisited. Journalof School Health, 69(3), 120-122.

Anonymous. (1999a). Calcium spurs sales of OJ and drinks. Prepared Foods, 168,33.

Anonymous. (1999b). Teenage wasteland? Not over food. Prepared Foods, 168,162.

Anonymous. (2000). Kitchen report on family eating habits: Family meals.Nutrition Today, 35(2), 60.

Baker, S.S., Cochran, W.J., Flores, C.A., Georgieff, M.K., Jacobson, M.S., Jaksic,T., et al. (1999). American Academy of Pediatrics. Committee on Nutrition.Calcium requirements of infants, children, and adolescents. Pediatrics, 104(5 Pt1), 1152-1157.

Ballew, C., Kuester, S., & Gillespie, C. (2000). Beverage choices affect adequacyof children’s nutrient intakes. Archives of Pediatric and Adolescent Medicine,154(11), 1148-1152.

Bingham, S.A., & Day, N.E. (1997). Using biochemical markers to assess thevalidity of prospective dietary assessment methods and the effect of energyadjustment. American Journal of Clinical Nutrition, 65(4 Suppl), 1130S-1137S.

Borrud, L., Enns, C.S., & Mickel, S. (1997). What we eat: USDA surveys foodconsumption changes. Community Nutrition Institute Newsletter, pp. 4-5.

Bowman, S. (1999). Diets of individuals based on energy intakes from addedsugars. Family Economics and Nutrition Review, 12(2), 31-38.

Calvo, M.S. (1994). The effects of high phosphorus intake on calciumhomeostasis. Advanced Nutrition Research, 9, 183-207.

Cavadini, C., Siega-Riz, A.M., & Popkin, B.M. (2000). US adolescent food intaketrends from 1965 to 1996. Western Journal of Medicine, 173(6), 378-383.

Dennison, B.A., Rockwell, H.L., & Baker, S.L. (1997). Excess fruit juiceconsumption by preschool-aged children is associated with short stature andobesity. Pediatrics, 99(1), 15-22.

Doucette, R.E., & Dwyer, J.T. (2000). Is fruit juice a “no-no” in children’s diets?Nutrition Review, 58(6), 180-183.

Federation of American Societies for Experimental Biology, Life SciencesResearch Office. (1995). Third Report on Nutrition Monitoring in the UnitedStates (Vols. 1 & 2). Washington, DC: U.S. Government Printing Office.


Goulding, A., Cannan, R., Williams, S.M., Gold, E.J., Taylor, R.W., & Lewis-Barned, N.J. (1998). Bone mineral density in girls with forearm fractures. Journalof Bone and Mineral Research, 13(1), 143-148.

Guthrie, J., & Morton, J. (2000). Food sources of added sweeteners in the diets ofAmericans. Journal of the American Dietetic Association, 100(1), 43-51.

Harnack, L., Stang, J., & Story, M. (1999). Soft drink consumption among USchildren and adolescents: Nutritional consequences. Journal of the AmericanDietetic Association, 99(4), 436-441.

Heaney, R.P., & Rafferty, K. (2001). Carbonated beverages and urinary calciumexcretion. American Journal of Clinical Nutrition, 74, 343-347.

Lee, R.D., & Nieman, D.C. (1996). Nutritional Assessment (2nd ed.). Chicago:Mosby.

National Academy of Sciences, Food and Nutrition Board, Institutes of Medicine.(1999). Dietary Reference Intakes for Ca, P, Mg, Vitamin D, and Fluoride.Retrieved August 2001, from http://www.nap.edu/books/0309063507/html/index.html.

National Family Opinion World Group. (1999). Available at http://www.nfow.com/NFOPanel.asp.

National Institutes of Health Consensus Conference. NIH Consensus DevelopmentPanel on Optimal Calcium Intake. (1994). Optimal calcium intake. Journal of theAmerican Medical Association, 272(24), 1942-1948.

National Soft Drink Association. (1999). Available at http://www.nsda.org.

The Pediatric Forum. (2001). Archives of Pediatric and Adolescent Medicine,155(2), 200-203.

Petridou, E., Karpathios, T., Dessypris, N., Simou, E., & Trichopoulos, D. (1997).The role of dairy products and non alcoholic beverages in bone fractures amongschoolage children. Scandinavian Journal of Social Medicine, 25(2), 119-125.

Russo, L. (1998). Seen & heard: Marketing to kids means a lot of looking &listening. Beverage World, 117(1666), 1-8.

Sfiligoj, E. (1998). Role reversal. Beverage World, 117(1659), 88-90.

Skinner, J.D., & Carruth, B.R. (2001). A longitudinal study of children’s juiceintake and growth: The juice controversy revisited. Journal of the AmericanDietetic Association, 101(4), 432-437.

Subar, A.F., Krebs-Smith, S.M., Cook, A., & Kahle, L.L. (1998). Dietary sourcesof nutrients among US children, 1989-1991. Pediatrics, 102(4 Pt 1), 913-923.

Tanasescu, M., Ferris, A.M., Himmelgreen, D.A., Rodriguez, N., & Perez-Escamilla, R. (2000). Biobehavioral factors are associated with obesity in PuertoRican children. Journal of Nutrition, 130(7), 1734-1742.

2002 Vol. 14 No. 2 79

Troiano, R.P., Briefel, R.R., Carroll, M.D., & Bialostosky, K. (2000). Energy andfat intakes of children and adolescents in the United States: Data from the NationalHealth and Nutrition Examination Surveys. American Journal of ClinicalNutrition, 72(5 Suppl), 1343S-1353S.

U.S. Department of Agriculture, Agricultural Research Service. (1997). Resultsfrom USDA’s 1996 Continuing Survey of Food Intakes by Individuals and 1996Diet and Health Knowledge Survey. Retrieved August 2001 (under “Releases”),from http://www.barc.usda.gov/bhnrc/foodsurvey/home.htm.

U.S. Department of Agriculture, & U.S. Department of Health and HumanServices. (1995). Nutrition and Your Health: Dietary Guidelines for Americans(4th ed.). Home and Garden Bulletin No. 232.

U.S. Department of Agriculture, & U.S. Department of Health and HumanServices. (2000). Nutrition and Your Health: Dietary Guidelines for Americans(5th ed.). Home and Garden Bulletin No. 232.

Weaver, C.M., Proulx, W.R., & Heaney, R. (1999). Choices for achievingadequate dietary calcium with a vegetarian diet. American Journal of ClinicalNutrition, 70(3 Suppl), 543S-548S.

Wyshak, G. (2000). Teenage girls, carbonated beverage consumption, and bonefractures. Archives of Pediatric and Adolescent Medicine, 154, 610-613.

Wyshak, G., & Frisch, R.E. (1994). Carbonated beverages, dietary calcium, thedietary calcium/phosphorus ratio, and bone fractures in girls and boys. Journal ofAdolescent Health, 15(3), 210-215.


Insight 23December 2000

The Role of Nuts in a Healthy Diet

Mark Lino, PhDKristin Marcoe, MBA, RDJulia M. Dinkins, PhDHazel Hiza, PhD, RDRajen Anand, PhD

U.S. Department of AgricultureCenter for Nutrition Policy and Promotion

While not a staple in the American diet,nuts are consumed by many Americans.This Nutrition Insight examines thecontribution nuts can make to a diet,consumption of nuts, characteristicsof people who eat nuts, and theassociation of nut consumption withdiet quality. Nuts are generally dividedinto two groups—tree nuts (almonds,pecans, etc.) and peanuts. Peanuts,technically a legume (a dried pea, bean,or lentil), are typically included in thenut group because they are used in amanner comparable to nuts and havea similar nutrient profile.

Contributions of Nutsto the Diet

Nut consumption can make beneficialcontributions to the diet but at a cost—increased calories. Recent researchindicates that frequency of nut con-sumption may have an inverse associa-tion with the risk of heart disease formen, women, and the elderly (Sabate,1999). Eating nuts also seems tolower serum cholesterol and favorablychanges a person’s lipoprotein profile(Sabate, 1993). Walnuts have specifi-cally been studied for their effect onserum lipids and blood pressure.Results have shown that incorporatinga moderate amount of walnuts into acholesterol-lowering diet decreasesserum total cholesterol levels andfavorably changes the lipoproteinprofile in healthy men (Sabate et al.,1993).

Growing evidence shows that nutshave bioactive constituents (like plantprotein, dietary fiber, and somemicronutrients) that elicit protectiveeffects on the heart. When subjects atetest diets including nuts, the choles-terol-lowering response was greaterthan predicted. This suggested thatconstituents of nuts, other than fattyacids, have additional cholesterol-lowering effects (Kris-Etherton et al.,1999). Findings from epidemiologicstudies suggest an inverse associationbetween death from stroke and intakeof the most concentrated food sourcesof vitamin E, such as nuts (Yochum etal., 2000). A review of the beneficialeffects of vegetarian foods, includingnuts, also has shown such foods leadto improved control of blood-glucoseconcentration, lower insulin require-ments, and better weight control fordiabetic patients (Segasothy et al.,1999).

Nut Consumption in theUnited States

Annual per capita consumption of nutsin the United States (based on fooddisappearance data) has ranged from7.0 to 8.9 pounds over the 1970-97period (fig. 1). Nut consumptiondecreased slightly between 1990 and1997 (8.9 to 8.5 pounds). Peanuts(which include peanuts in peanut butterand candy) accounted for most of thisnut consumption—68 percent of totalnut consumption in 1997. Tree nuts

Research Briefs

2002 Vol. 14 No. 2 81

accounted for the minority share of percapita nut consumption. In 1997, themost commonly consumed tree nutswere almonds (19 percent of per capitatree nut consumption), coconuts (18percent), pecans (17 percent), andwalnuts (17 percent). As a percentageof total annual per capita consumptionof food, nuts accounted for a very smallshare, compared with other foods. Forexample, the per capita consumption ofmeat, poultry, and fish was 190 poundsin 1997, compared with 8.5 pounds forall nuts (Putnam et al., 1999).

Nut Eaters and the Qualityof Their Diet

Who in the United States eats nuts?And what is the quality of their diet?To answer these questions, we useddata from the Market Research Corpo-ration of America (MRCA) InformationServices. We used information from6,928 people for the 1992-94 period;these data were weighted to provide

2.2 2.22.9 2.7

1970 1980 1990 19970

1

2

3

4

5

6

7

8

9

10

PeanutsTree nuts

5.54.8

6.05.8

Figure 1. Annual per capita consumption of nuts

Pounds

Source: U.S. Department of Agriculture, Economic Research Service.

population estimates. MRCA collectedinformation on people’s consumptionof nuts based on detailed diaries offoods eaten over a 14-day period.“Nut eaters” were defined as peoplewho consumed any type of nut over a14-day period, and “non-nut eaters”were defined as people who did notconsume any type of nut over this sameperiod. Forty-one percent of peoplewere “nut eaters,” and 59 percent were“non-nut eaters.”

Of the five characteristics examined(gender, age, income, race, and regionof residence), age and race of nut eatersand non-nut eaters were significantlydifferent (table 1). Compared withnon-nut eaters, a significantly higherpercentage of nut eaters were youngerand were White. Among nut eaters, 37percent were under age 19, comparedwith 25 percent of non-nut eaters. Inaddition, among nut eaters, 90 percentwere White, compared with 86 percentof non-nut eaters.

To answer the question of whether nuteaters have a better diet than do non-nuteaters, we used a modified version ofthe Healthy Eating Index (HEI). Thismodified version uses 9 of the original10 HEI components. Components 1-5measure the degree to which a person’sdiet conforms to serving recommenda-tions of the Food Guide Pyramid foodgroups: Grains (bread, cereal, rice, andpasta), vegetables, fruits, milk (milk,yogurt, and cheese), and meat (meat,poultry, fish, dry beans, eggs, and nuts).As a percentage of total intake of foodenergy, component 6 measures con-sumption of total fat; component 7,saturated fat. Component 8 measurestotal cholesterol intake; component 9,sodium intake. The score for each com-ponent ranges from zero to 10, withhigher component scores indicatingintakes that are closer to recommenda-tions. The MRCA data set does notprovide enough information to calculatethe variety of a person’s diet (compo-nent 10 of the original HEI), so varietywas not calculated. All total HEI scoreson the modified version were adjustedto a 100-point scale. Scores greater than80 imply a good diet; between 51 and80, a diet that needs improvement; andless than 51, a poor diet.

For nut eaters, the mean score on themodified HEI was slightly, but signifi-cantly, higher than the score for non-nut eaters (60.8 vs. 56.9). Both groups,however, had total scores that indicatedtheir diet needed improvement. Theaverage daily caloric intake of nuteaters also was significantly higher—about 10 percent—than that of non-nuteaters. On the individual componentsof the HEI, nut eaters, compared withnon-nut eaters, had significantly higherscores for grains, fruits, milk, fat,saturated fat, and cholesterol. Thehigher HEI fat score—indicating nuteaters consume less fat as a percentageof total calories—may seem surprisingbecause nuts contain fat. It appears thatnut eaters consume less fat from other


Table 1. Characteristics of nut eaters and non-nut eaters, 1992-94

Characteristic Nut eaters Non-nut eaters

PercentGender Male 47 46 Female 53 54Age (years)* <19 37 25 19 - 51 36 51 >51 27 24Income 200% of poverty or less 45 42 More than 200% of poverty 55 58Race* White 90 86 Non-White 10 14Region Northeast 19 20 South 25 25 Midwest 33 35 West 23 20

*Significant at .05 level, based on unweighted data.

foods in their diet, compared with non-nut eaters, or given their caloric levelis higher, they consume less fat as apercentage of total energy. Non-nuteaters had significantly higher scoresfor vegetables and sodium that did nuteaters. The meat score for the twogroups was not significantly different.

Other factors (such as age and race)may influence the modified HEIscores of nut eaters and non-nut eaters.However, even when using multivariateanalytic procedures to control for allfive characteristics previously exam-ined, we found that the modified HEIscore for nut eaters was significantlyhigher than the score for non-nut eaters.

Conclusion

Although nut consumption is low,compared with other protein sourcessuch as meat and poultry, nuts providemany of the same nutrients to the dietand have potential health benefits. Asignificantly higher percentage of nuteaters than non-nut eaters were youngerand were White. In addition, comparedwith non-nut eaters, nut eaters had aslightly better diet, albeit one thatneeded improvement.

References

Kris-Etherton, P.M., Yu-Poth, S.,Sabate, J., Ratcliffe,H.E., Zhao, G., &Etherton, T.D. (1999). Nuts and theirbioactive constituents: Effects on serumlipids and other factors that affectdisease risk. The American Journal ofClinical Nutrition, 70(3S), 504S-511S.

Putnam, J.J., & Allshouse, J.E. (1999).Food Consumption, Prices, andExpenditures, 1970-97. Food andRural Economics Division, EconomicResearch Service, U.S. Department ofAgriculture. Statistical Bulletin No.965.

Sabate, J. (1999). Nut consumption,vegetarian diets, ischemic heart diseaserisk, and all-cause mortality: Evidencefrom epidemiologic studies. TheAmerican Journal of Clinical Nutrition,70(3S), 500S-503S.

Sabate, J. (1993). Does nut consump-tion protect against ischaemic heartdisease? European Journal of ClinicalNutrition, 47(1), S71-S75.

Sabate, J., Fraser, G.E., Burke, K.,Knutsen, S.F., Bennett, H., & Lindsted,K.D. 1993. Effects of walnuts on serumlipid levels and blood pressure innormal men. New England Journal ofMedicine, 328(9), 603-607.

Segasothy, M. & Phillips, P.A. (1999).Vegetarian diet: Panacea for modernlifestyle diseases? Quarterly Journal ofMedicine, 92(9), 531-544.

Yochum, L.A., Folsom, A.R., & Kushi,L.H. (2000). Intake of antioxidantvitamins and risk of death from strokein postmenopausal women. TheAmerican Journal of Clinical Nutrition,72(2), 476-483.

2002 Vol. 14 No. 2 83

Insight 24May 2001

Food Trade-Offs: Choosing Howto Balance the Diet

Julia M. Dinkins, PhDMark Lino, PhD

U.S. Department of AgricultureCenter for Nutrition Policy and Promotion

People often make “food trade-offs”to balance their diet. For example, aperson may choose to eat lowfat frozenyogurt rather than regular ice cream.The trade-offs that people make, whomakes these trade-offs, and how thesetrade-offs affect their diet are of interestto nutrition educators. This Insightexamines in-depth the food trade-offspeople make. From a list of 14 foods,people indicated whether they usedany of five trade-offs to balance thenutrition in their diet. Most peoplemade at least one food trade-off—typically limiting how often they ate afood. And those who made food trade-offs had a more healthful diet, com-pared with those who made none.

Source of Data

We used data from Market ResearchCorporation of America (MRCA)Information Services for this analysis.MRCA conducts a National ConsumerPanel. Households are selected basedon demographic criteria matched to theU.S. Census. We used information from5,787 adults in these households for the1992-94 period. All data were weightedto the population.

These adults were asked what foodtrade-offs they typically made tobalance the nutrition in their dietwhen they ate 14 foods: red meat,eggs, cheese, breads, margarine, saladdressing, chips, pretzels, sugared softdrinks, ice cream, cakes, cookies, sweetrolls/donuts, and sugar. These foods aretypically characterized as being high in

fat, cholesterol, sodium, and/orcalories. MRCA compiled a list of foodtrade-offs people may use and verifiedthe list in the pretesting phase of datacollection. People could select one ormore of five trade-offs they made withregards to each food: limit how ofteneaten, limit portion size, give up someother food (e.g., consume ice cream,but do not consume cake because ofthis), substitute a healthier version offood (e.g., consume extra lean redmeat rather than regular red meat), orsubstitute a food item with another fooditem at other times (e.g., consume chipsas a snack today, but consume an appleas a snack tomorrow). So, for eachfood, people could make no trade-offor could make up to five trade-offs. Themaximum number of trade-offs theycould make for all 14 foods was 60.

Most People Make FoodTrade-Offs

Most people (86 percent) reportedmaking at least one food trade-off(fig. 1). Forty-one percent reportedmaking 1 to 10 food trade-offs, and45 percent reported making 11 or moretrade-offs. The trade-off most oftenreported was limiting how often aparticular food was eaten: 82 percentof the people said they made this trade-off for at least one of the 14 foods(fig. 2). The next trade-off that wasmost frequently used was limiting theportion size: 53 percent of the peoplesaid they made this trade-off for at leastone of the 14 foods. The least likelytrade-off strategies people used were


substituting a healthier version of thefood (18 percent), substituting withanother food at other times (10 per-cent), and giving up some other food(8 percent).

Some People More LikelyThan Others to Make FoodTrade-Offs

Personal characteristics influence one’slikelihood to make food trade-offs. Ofthe characteristics examined, gender,age, race, and education were signifi-cantly different for people making foodtrade-offs (table). Compared withmales, females were much more likelyto make a food trade-off. Ninety-twopercent of females reported making atleast one food trade-off; 79 percent ofmales reported making a trade-off.Older adults were more likely to make afood trade-off than were their youngercounterparts: 90 percent of people ages51 and over versus 76 percent of peopleages 18 to 30 made a trade-off.

Non-whites were more likely to make afood trade-off than were whites (91 vs.85 percent). People with more educa-tion also were more likely to make afood trade-off. Ninety percent of adultswith more than a high school diplomamade a food trade-off; 82 percent ofadults with a high school diploma orless did so.

People Who Make FoodTrade-Offs Have a“Better Diet”

To answer the question of whetherpeople who use a food trade-off have a“better diet” than those who do not, weused a modified version of the HealthyEating Index (HEI). This version uses9 of the original 10 HEI components.

Figure 1. People are likely to use tradeoffs to balance their diet

41%

14%

45%

0 Trade-offs 1-10 Trade-offs 11+ Trade-offs

Limit how often eaten

Limit portion size

Substitute healthier version

Substitute with another food

Give up some other food

0

20

40

60

80

100

Per

cen

t o

f P

eop

le

Figure 2. People are most likely to limit how often certain foods are eaten

82%

18%

53%

8%10%

2002 Vol. 14 No. 2 85

Table. Food trade-offs areinfluenced by people’scharacteristics

Made Did not makeCharacteristic trade-offs trade-offs

PercentGender*

Male 79 21Female 92 8

Age*18-30 76 2431-50 87 1351+ 90 10

Income< $21,000 85 15$21,000-$40,000 85 15$41,000+ 89 11

Race*White 85 15Non-white 91 9

Education*12 years of education or less 82 18More than 12 years of education 90 10

* Significant at .05 level, based on unweighted data.

Components 1-5 measure the degreeto which a person’s diet conforms toserving recommendations of the FoodGuide Pyramid food groups: Grains(bread, cereal, rice, and pasta), veg-etables, fruits, milk (milk, yogurt, andcheese), and meat (meat, poultry, fish,dry beans, eggs, and nuts). As apercentage of total intake of foodenergy, component 6 measures con-sumption of total fat; component 7,saturated fat. Component 8 measurestotal cholesterol intake; component 9,sodium intake. The score for eachcomponent ranges from zero to 10.The MRCA data set does not provideenough information to calculate thevariety of a person’s diet (component10 of the original HEI), so variety wasnot calculated. All total HEI scores onthe modified version were adjusted to a

100-point scale. Scores greater than 80imply a “good diet”; between 51 and80, a diet that “needs improvement”;and less than 51, a “poor diet.”

Adults who reported making 11 ormore food trade-offs for the 14 foodshad a significantly higher HEI score(60.0) than was the case for adults whoreported making 1 to 10 trade-offs(57.0) and for adults who reportedmaking no food trade-offs (53.5).Additional analysis showed that adultswho made at least one food trade-off,compared with those who made notrade-offs, had significantly higherHEI component scores for grains,fruit, vegetables, fat, saturated fat,and cholesterol. These higher scoresindicate that people had intakes thatwere closer to recommendations forthese HEI components.

Conclusion

Most people report making food trade-offs to balance the nutrition in theirdiet. The most common trade-offs arelimiting how often a food is eaten andlimiting portion size. People who makesuch trade-offs have a better diet thanthose who do not (although both groupshad average HEI scores indi-cating adiet that “needed improvement”).Nutrition educators may find thisinformation on food trade-offs helpfulin the design of realistic nutritionintervention programs and individualcounseling and as a technique forencouraging healthful eating.


From the Food and NutritionService; Office of Analysis,Nutrition, and Evaluation

The School BreakfastPilot Project

The School Breakfast Program,authorized by the Child Nutrition Actof 1966, started as a pilot program toprovide funding for school breakfasts tochildren in poor areas and areas wherethey had to travel a great distance toschool. The intent was to provide anutritious breakfast to children whomight otherwise not receive one. Theimportance of a nutritious breakfast issupported by the growing body ofevidence that has linked it to improveddietary status and enhanced schoolperformance. Hence, many observershave urged that school breakfastsbecome more available.

Despite an increase in the number ofschools offering the School BreakfastProgram, the percentage of studentswho eat school breakfasts is consider-ably lower than the percentage who eatschool lunch. Those eating schoolbreakfasts are more likely to be poorand qualify for free or reduced-pricebreakfasts. Some people believe that auniversal-free program would result inmore children consuming a nutritiousbreakfast and beginning the school daywith proper nutrition and ready to learn.

Within this context, Congress passedSection 109 of the William F. GoodlingChild Nutrition Act of 1998 (P.L. 105-336), which authorized the Secretaryof Agriculture, through the U.S.Department of Agriculture’s Food andNutrition Service (FNS), to conduct a3-year pilot project beginning in the

2000-2001 school year that providedfree school breakfasts to all studentsregardless of family income.

Six of the 386 school districts wereselected from those that applied toparticipate in the pilot project:• Boise, Idaho: Independent School

District of Boise• Columbiana, Alabama: Shelby

County Board of Education• Gulfport, Mississippi: Harrison

County School District• Phoenix, Arizona: Washington

Elementary School District• Santa Rosa, California: Santa Rosa

City Schools• Wichita, Kansas: Wichita Public

Schools

A total of 143 elementary school unitsfrom the participating districts weregrouped into matched pairs on thebasis of several demographic variables.One school unit in each pair wasrandomly assigned to the treatmentgroup (universal-free school breakfast)or the control group (regular schoolbreakfast). Within each treatment andcontrol school unit, about 30 studentswere selected for the evaluation. Thetotal student sample size was 4,290(2,190 treatment and 2,100 control).Data were collected from students,parents, teachers, school district staff,and school records during spring 2001.

FNS’s evaluation of the 3-year projectconsists of an implementation study andan impact study. The implementationstudy describes how the schools chooseto implement universal-free schoolbreakfast, assesses the effect ofstudents’ participation in the universal-free school breakfast on administrativerequirements and costs at the schooland federal levels, and assesses nutrient

composition of the school breakfasts.The impact study assesses the effectsof universal-free breakfast on a broadrange of student outcomes: includingschool breakfast participation, breakfastconsumption patterns, dietary intake,food security status, school attendanceand tardiness, child health, cognitivefunctioning, classroom behavior andattentiveness, and academic achieve-ment.

An interim report that provides resultsfrom the first year of the pilot waspublished in October 2002; a finalreport, in summer 2004. Additionalinformation on the School BreakfastPilot Project can be found at http://www.fns.usda.gov/OANE/MENU/sbppilot/sbpnotice.htm.

National School LunchProgram Application/Verification Pilot Projects

This project responds to FNS’s growingconcern about program integrity issuesassociated with the current system bywhich School Food Authoritiesdetermine eligibility for free andreduced-price school meals. Twenty-two School Food Authorities across16 States began testing pilot proceduresin 2000-2001 to determine and verifychildren’s eligibility for free andreduced-price school meals. Threemodels are being tested over a 3-yearperiod to determine eligibility forthese school meals: Up-FrontDocumentation, Graduated Verifica-tion, and Verify Direct Certification.Evaluation of the models will permitFNS to explore a variety of optionsto improve the targeting of free andreduced-price meals to income-eligiblechildren.

Research and Evaluation Activities in USDA

Regular Items

2002 Vol. 14 No. 2 87

The Year One Report will providea descriptive statistical analysis of dataFNS has collected from 19 SchoolFood Authorities in 14 States thatoperated the pilot project in fiscalyear 2000-2001. FNS is collectingadministrative data from the participat-ing pilot sites regarding several keyprogram statistics such as percentageof children approved for free meals andreduced-price meals, number of mealsserved at free and reduced-price levels,and results of graduated and directcertification verification. Data reportedby School Food Authorities to FNS forthis purpose represent information allSchool Food Authorities are required tomaintain for reporting purposes. Thesedata allow FNS to compare changes inkey program statistics in these SchoolFood Authorities between the first pilotschool year and the 2 pre-pilot years.

FNS anticipates conducting a morerigorous evaluation of the Up-FrontDocumentation and GraduatedVerification pilot projects. Central tothis evaluation are a comparison ofthe free and reduced-price approvalstatus and an independent collectionof income information for householdsin the pilot School Food Authorities.Hence, FNS will be able to address abroader range of issues than is possiblethrough exclusive reliance upon SchoolFood Authorities’ administrative data.Results from this evaluation arescheduled for release in fiscal year2003.

Characteristics of Food StampHouseholds

During each month of fiscal year 2000,about 17.2 million people living in 7.3million U.S. households received foodstamps.

• Most food stamp recipients arechildren or the elderly. Overhalf (51 percent) are children; 10percent, age 60 or older. Working-age women represent 27 percentof the caseload; working-age men,11 percent.

• Most food stamp households donot receive cash welfare benefits.Nearly a third (32 percent) receiveSupplemental Security Income; one-quarter (25 percent), TemporaryAssistance for Needy Familiesbenefits; and one-quarter (25percent), Social Security benefits.Eight percent have no cash incomeof any kind.

• Many food stamp recipients work.Over one-fourth (27 percent) offood stamp households haveearnings, the primary source oftheir income.

• Food stamp households have littleincome. Only 11 percent are abovethe poverty line, while 33 percenthave incomes at or below half thepoverty line. The typical food stamphouseholds have gross income of$620 per month and receive amonthly food stamp benefit of $158.Food stamps account for over one-fifth of monthly funds (cash incomeplus food stamps) available to atypical household.

• Food stamp households possessfew resources. The average foodstamp household possesses onlyabout $156 in countable resources(including the nonexcludedportion of vehicles and the entirevalue of checking and savingsaccounts and other savings).

• Most food stamp households aresmall. The average food stamphousehold size is 2.3 but variesconsiderably by householdcomposition. Households withchildren are relatively large,averaging 3.4 members. Householdswith elderly members tend to besmaller, averaging 1.3 members.

For more information on theCharacteristics of Food StampHouseholds: Fiscal Year 2000, visitwww.fns.usda.gov/OANE/MENU/published/FSP/Participation.htm.

Rates of Food Stamp ProgramParticipation

An important measure of a program’sperformance is its ability to reach itstarget population. For over 15 years,the national food stamp participationrate—the percentage of eligiblepeople who participate actively inthe program—has been a standard forassessing the program’s performance.Trends in Food Stamp ProgramParticipation Rates: 1994-1999provides the latest information on theprogram’s participation rates, based ondata from the Census Bureau’s CurrentPopulation Survey and administrativedata.

Overall Trends: Participation rates fellby 17 percentage points between 1994and 1999. While the number of eligibleindividuals fell by 16 percent, thenumber of participating individuals fellby 35 percent. The overall participationrate among people eligible for benefitswas 57 percent in 1999, down from ahigh of 74 percent in 1994.

Trends Among Subgroups:Participation rates among childrendeclined each year between 1994 and1999, dropping from nearly 90 percentin 1994 to 68 percent by 1999. Thedecrease occurred between bothpreschool children and school-agedchildren. Participation rates fell bothfor individuals in single-parenthouseholds and for those in married-couple families with children.

By contrast, participation rates between1994 and 1999 were fairly stable butconstantly low among the elderly:about 30 percent of eligible seniors


participated. Among the disabled,about half of the eligible participantsparticipated between 1994 and 1997.However, their participation rate rose to61 percent by 1999. Participation ratesdeclined each year among individualsliving in households without anyworkers, dropping from 85 percent in1994 to 70 percent in 1999. Amongworking-poor households, participationrates among those who were eligiblewere stable between 1994 and 1997 butfell between 1997 and 1999: an averageof 53 and 48 percent, respectively.

Contrary to expectations, participationrates have fallen most rapidly amongthose living in households that qualifyfor the largest benefits (over half ofthe maximum allotment). In 1994,98 percent of these individualsparticipated; by 1999, 78 percentparticipated. These householdsgenerally have the lowest income,relative to their household size.

Assessment of ComputerMatching in the Food StampProgram

Computer matching is a powerfulmanagement tool, which has increasedthe integrity and efficiency of the FoodStamp Program. This study examinedhow States are currently using orplanning to use computer-matchingstrategies for error reduction.

Since 1991, States, in general, havesignificantly increased their use ofcomputer matching to enhance programintegrity. Advances in computertechnology facilitated development ofstrategies to make fraud more difficultto commit and easier to detect. Tech-nological advances, particularly thegrowth in communications networks,greatly increased States’ matchingcapability, which led to more rapid

responses from external databases.For more information on Assessment ofComputer Matching in the Food StampProgram, visit http://www.fns.usda.gov/OANE/MENU/Published/FSP/NewReleases.htm.

FNS Planning andPerformance Measurement

FNS is responsible for managing theAgency’s strategic and operationalplanning systems. The Office ofAnalysis, Nutrition, and Evaluation(OANE) coordinates the Agency’scontributions to the Department’sstrategic plan, performance plans,and performance reports, which arerequired by the Government Perfor-mance and Results Act (GPRA, P.L.103-62). OANE also manages aninternal process intended to align theAgency’s operational priorities with thegoals of the strategic and annual plans,as well as other key program and policyobjectives.

One critical part of OANE’s responsi-bilities is to identify and improvemeasures of program performancethat link to FNS’s strategic goals andobjectives. The current FNS strategicplan, released in September 2000,includes two major goals and fiveobjectives that cut across program linesand represent the shared purposes andintended outcomes of Federal nutritionassistance:

1. Improved Nutrition of Children and Low-Income People

• Improved Food Security

• Program Participants MakeHealthy Food Choices

• Improved Nutritional Qualityof Meals, Food Packages,Commodities, and Other ProgramBenefits

2. Improved Stewardship of Federal Funds

• Improved Benefit Accuracy andReduced Fraud

• Improved Efficiency of ProgramAdministration

For each objective, FNS has sought todevelop performance measures thatcan be used to measure programperformance and target improvement.For example, the Agency measuresprogress toward its objective to“improve food security” throughthe USDA food security measure,specifically focusing on reducing theprevalence of hunger among childrenand low-income people. FNS supple-ments the hunger measure, a subscaleof the food security measure, withmeasures of “coverage”—the rate ofparticipation among eligible people—for the major Federal nutritionassistance programs. Juxtaposing theprevalence of hunger among childrenand low-income people with rates ofparticipation helps provide a basis forevaluating the relationship between theprogram participation output and theanti-hunger outcomes the programs areintended to influence.

Similarly, for its objective that“program participants make healthyfood choices,” FNS uses the HealthyEating Index, a measure of diet qualitydeveloped by USDA’s Center forNutrition Policy and Promotion, to settargets to improve the nutrition statusof children and low-income people.FNS also seeks to increase the rate ofbreastfeeding—the feeding practicejudged by nutrition and health expertsto be the best for infants—amongmothers and infants participating in theWomen, Infants, and Children (WIC)program.

2002 Vol. 14 No. 2 89

A number of significant challengesexist in developing performancemeasures for FNS programs; three areparticularly significant:

1. Program Structure: Most Federalnutrition assistance programsoperate as partnerships between theFederal Government and State andlocal authorities, so the link betweenAgency work performance andprogram effectiveness is indirect.

2. “Context” of ProgramPerformance: As with many otherGovernment programs, the out-comes that the programs areintended to influence, such asreduction in hunger and improve-ment in diet quality, are alsoinfluenced by many factors beyondthe Agency’s control.

3. Data Limitations: Data sourcesfor many areas of performance arelimited by constraints on Agencyand program resources, includingthe need to reduce the burden onlocal program operators.

To deal with these challenges, FNScontinually identifies and developsimproved performance measures of itsprograms and operations in a number ofareas. Working with program offices,OANE seeks to gain insight in specificaspects of program performance byleveraging internal Agency operationaldata, the results of oversight and reviewwork; improving collection of programdata; and developing stand-aloneevaluation work.

Improvement in planning andperformance measurement arenecessarily incremental; over time,however, the broad framework of thestrategic plan helps to ensure that theteam and the Agency focus their effortson the most critical areas of perfor-mance. These efforts should help theAgency gain new insights and developand test innovative strategies thatimprove program effectiveness.


Food Supply Nutrients and Dietary Guidance, 1970-99

Using data on per person consumption and information on nutrient composition, USDA’s Center for Nutrition Policy andPromotion calculates the nutrient content of the food supply. Per person consumption for each commodity is multiplied bythe amount of food energy and each of 27 nutrients and dietary components in the edible portion of the food. Results foreach nutrient from all foods are totaled and converted to amount per person per day. Nutrients added commercially to certaincommodities (i.e., through fortification and enrichment) are also included in the nutrient content of the food supply. Foodsupply data represent the disappearance of food into the marketing system; therefore, per person consumption and nutrientestimates typically overstate the amount of food and nutrients people actually ingest.

Food supply providing morecarbohydrates, protein, and fat

In 1999 the food supply provided500 grams of carbohydrate, 111 gramsof protein, and 164 grams of total fatper person per day. This was anincrease from 1970, when the foodsupply provided 389 grams ofcarbohydrate, 96 grams of protein,and 151 grams of total fat per personper day.

Federal Studies

Folate in food supply increasedsharply after 1998

Folate reduces the risk of some seriousbirth defects when consumed beforeand during pregnancy. Mandated folatefortification of flours and cereals in1998 increased the folate level in theU.S. food supply. In 1999 the level offolate was 641 micrograms per personper day, an increase of more than 130percent from the 1970 level of 278micrograms per person per day.

Percentage increase in carbohydrate, protein, and total fat in U.S. food supply(per person per day) from 1970 to 1999

Folate in U.S. food supply (per person per day) over time

Carbohydrate

Protein

Total fat

0 5 10 15 20 25 30 35

29%

16%

9%

1970 75 80 85 90 95 99

200

300

400

500

600

700

Mic

rogr

ams

2002 Vol. 14 No. 2 91

Sources of calcium in the U.S. food supply

Sources of iron in the U.S. food supply

More calcium coming from lowfatand skim milk and less from wholemilk

Calcium levels have generallyincreased in the food supply, from 930milligrams in 1970 to 990 milligramsper person per day in 1999. Lowfatand skim milk provided 11 percent ofcalcium in the food supply in 1970 and21 percent in 1999. On the other hand,whole milk provided 35 percent ofcalcium in the food supply in 1970 and11 percent in 1999. This shift supportsdietary guidance that recommends dietslow in saturated fats.

More iron coming from grains andbreakfast cereals and less frommeat, poultry, fish, and meatalternates

Iron levels in the food supply increasedfrom 15.3 milligrams per person perday in 1970 to 23.6 milligrams perperson per day in 1999. Grains andbreakfast cereals provided 35 percentof iron in the food supply in 1970and 53 percent in 1999. Meat, poultry,fish, and meat alternates provided 32percent of iron in the food supply in1970 and 23 percent in 1999. Enrichedgrains and fortified ready-to-eatbreakfast cereals contributed to theincrease in grains and cereals as asource of iron.

Source: Gerrior, S. & Bente, L. (2001). Food supply nutrients and dietary guidance, 1970-99. FoodReview, 24(3), 39-46.

1970 1999

1970 1999

2%

23%

53%

2%3%

7%

10%

Vegetables

Whole milk

Cheese

Lowfat and skim milk

Other foods

6%

18%

35%

19%

11%

11%

Other dairy

7%

25%

21%21%

15%

11%

2%

Grains andbreakfastcereals

Meat, poultry, fish,and meat alternates

Vegetables

Other foods

Eggs

FruitsDairyproducts

35%

3%4%

10%

14%

32%


Poverty rates by residence over the 1990’s

Distribution of the poor by residence, 1999

Rural Poverty

About 7.4 million people who live in rural (nonmetro) areas were poor in 1999. Based on data from the U.S. CensusBureau’s Current Population Survey (March Supplement), the rate of rural poverty for 1999 was 14.2 percent. Rural povertyrates were not under 15 percent for 2 consecutive years at any other time in the 1980’s or 1990’s. The rural poverty rate for1999 was the lowest it has been since 1979; similarly, the national and urban poverty rates in 1999 were also at their lowestlevels since 1979. These relatively low levels of poverty coincided with the economic boom in the United States.

Rural and urban poverty ratesdeclined in the 1990’s

Poverty rates declined in both ruraland urban areas in the 1990’s, withrural rates higher than urban rates.From 1998 to 1999, the urban povertyrate declined at a greater pace (from12.3 to 11.2 percent) than did therural poverty rate (from 14.3 to 14.2percent). This widened the gap in rural-urban poverty, from the average 2.6percentage points of the 1990’s to 3percentage points between 1998 and1999.

Lower percentage of the extremelypoor live in rural areas

Of the poor in 1999, 35.3 percent ofthe rural poor were extremely poorversus 40.5 percent of the urban poor.Extreme poverty is defined as havingan income less than half the povertyline. Similarly, 38 percent of the ruralpoor versus 32.1 percent of the urbanpoor had incomes between 75 and 100percent of the poverty line.

1990 93 96 9910

11

12

13

14

15

16

17

18

Nonmetro Metro

Nonmetro Metro

35.3% 38%

26.7%

40.5%

32.1%

27.4%

Income 75-100% of poverty line

Income 50-74% of poverty line

Income under 50% of poverty line

2002 Vol. 14 No. 2 93

Poverty rates by region and residence, 1999

Poverty rates by age and residence, 1999

Rural poverty rates highest in South

Poverty rates differed by U.S. region.In 1999 the South had the highest rateof rural poverty (17 percent) and theNortheast the lowest (9.4 percent).The West had the highest rate of urbanpoverty (12.2 percent); the Midwest,the lowest (9.2 percent).

Rural poverty rates highestfor children

The 1999 rural poverty rate forchildren was 19.8 percent, resultingin 2.7 million children being affected.This poverty rate was 7.4 percentagepoints greater than the rate for adultsand 8.1 percentage points greater thanthe rate for elderly persons living inrural areas. For all age categories, ruralpoverty rates were higher than urbanrates in 1999.

Northeast Midwest South West0

5

10

15

20

Nonmetro Metro

11.1% 11.5%

9.2%

17.0%

11.9%

15.4%

12.2%

9.4%

0-17 years 18-64 years 65 years and older0

5

10

15

20

25

Nonmetro Metro

19.8%

16.2%

12.4%

9.5%11.7%

9.1%

Source: Joliffe, D. (2002). Rural poverty rate stayed under 15 percent in 1999. Rural America, 16(4), 39-41.


Pay status of people doing job-related work at home, 2001

Pay status of people in selected occupations doing job-related work at home,2001

Work at Home in 2001

In May 2001, 19.8 million people age 16 and over—accounting for 15 percent of total employment—usually performedsome work at home as part of their primary job. People who usually work at home are defined as those who work at homeat least once per week as part of their primary job. These findings are from a special supplement to the May 2001 CurrentPopulation Survey, a survey of about 50,000 households conducted by the U.S. Census Bureau for the Bureau of LaborStatistics. Respondents to the supplement answered questions about work schedules, job-related work at home, and relatedtopics.

Half of those who worked at homewere not directly compensated

Slightly more than half of those whousually worked at home were wageand salary workers who took workhome from the job and worked on anunpaid basis. Another 17 percent hada formal arrangement with theiremployer to be paid for their at-homework. The remainder who worked athome were self-employed.

More managers and professionalsworked at home on an unpaid basis

Of people in managerial andprofessional occupations usuallydoing job-related work at home,most (64 percent) were unpaid. Thiscontrasted with the 41 percent ofpeople in technical, sales, andadministrative support and the19 percent of people in serviceoccupations usually doing unpaidjob-related work at home.

Self-employed Unpaid work at homePaid work at home

25%25%

Managerial andprofessional

Technical, sales, andadministrative support

Service0

20

40

60

80

100

14%25% 25%

22%

34%

56%

64%

41%

19%

17%

30%

53%

Paid work at home Self-employed Unpaid work at home

2002 Vol. 14 No. 2 95

Average hours usually worked at home

Reasons for working at home when not paid

People who worked at home on apaid basis worked more hours

The average time worked at home forpeople expressly paid for this workwas 18 hours; 1 in 6 of these peopleput in 35 hours or more at home.The average time worked at home forpeople without a formal arrangementto be paid for this work was 6.8 hours;23 percent worked 8 or more hours athome.

Most people doing unpaid work athome were trying to finish or catchup on work

When asked the reason for theirworking at home without pay, 60percent of those who usually work athome stated they did so to finish orcatch up on work; 33 percent said theydid so because of the nature of thejob; the remainder gave other reasons(e.g., coordinate work schedule withpersonal or family needs, reducecommuting time, or reduce expenses).

Unpaid work at home

Paid work at home

0 5 10 15 20

6.8

18.0

Source: Bureau of Labor Statisics. Work at home in 2001. Retrieved at: http://www.bls.gov/cps, March 27, 2002.

Other

60%

7%

33%

Finish/catch up on work Nature of job


Leisure-Time Physical Activity Among Adults, 1997-98

This report, based on data from the 1997-98 National Health Interview Survey (NHIS), presents selected prevalenceestimates for leisure-time physical activity among U.S. adults. Computer-assisted personal interviews were used to collectdata from the U.S. civilian noninstitutionalized population. A total of 68,556 interviews was completed by a sample of adultsaged 18 years and over, resulting in an overall response rate of 77.2 percent. NHIS questions on leisure-time physicalactivity, first asked in 1997, consisted of frequency and duration of light-moderate activity, frequency and duration ofvigorous activity, and frequency of strengthening activity. Statistics were age-adjusted to the 2000 projected U.S. population.Adults classified as engaging in at least some activity were those who engaged in any light-moderate or vigorous leisure-time physical activity for at least 10 minutes, regardless of frequency.

Six in ten adults engaged in somephysical activity

Sixty-two percent of adults engagedin some leisure-time physical activityduring the year. Thirty-one percentengaged in regular physical activity(light-moderate activity 5 times ormore per week for 30 minutes or moreeach time or vigorous activity 3 timesor more per week for 20 minutes ormore each time). Twenty-three percentengaged in strengthening activities(e.g., lifting weights or calisthenics)designed to strengthen muscles.

Adults engaged in physical activity

Men more likely to engage in somephysical activity

Men were more likely than women toengage in some leisure-time physicalactivity (65 vs. 59 percent). Men alsowere slightly more likely than womento engage in light-moderate or vigorousactivity or both at least 5 times perweek, as well as strengtheningactivities.

Men and women engaged in some physical activity

0 10 20 30 40 50 60 70

23%

31%

62%

Strengthening activity

Regular physical activity

Some physical activity

Men Women0

10

20

30

40

50

60

70

65%

59%

2002 Vol. 14 No. 2 97

Physical activity decreased with age

Adults 18-24 years old were almosttwice as likely as adults 75 yearsand over to engage in some physicalactivity (70 vs. 39 percent). Adults inthe younger age groups also were morelikely than those in the older agegroups to engage in light-moderate,vigorous, and strengthening physicalactivity.

Adults engaged in some physical activity, by age

Whites more likely to engage inphysical activity

White non-Hispanic adults (66percent) and Asian/Pacific Islandernon-Hispanic adults (62 percent) weremore likely than Black non-Hispanicadults (50 percent) and Hispanic adults(47 percent) to engage in some leisure-time physical activity. Engaging inlight-moderate physical activity at least5 times per week, as well as vigorousphysical activity, at least 5 times perweek was more prevalent among Whitenon-Hispanic adults than among otherrace-ethnic groups.

Adults engaged in some physical activity, by race-ethnicity

0

10

20

30

40

50

60

70

Whitenon-Hispanic

Blacknon-Hispanic

Asian-PacificIslander

non-Hispanic

Hispanic

66%

50%

62%

47%

18-24 25-44 45-64 65-74 75+

Years

0

10

20

30

40

50

60

70

80

70%67%

61%

52%

39%

Source: Schoenborn, C.A. & Barnes, P.M. (2002). Leisure-time physical activity among adults: United States, 1997-98.Advance Data from Vital and Health Statistics. National Center for Health Statistics, Centers for Disease Control andPrevention, Department of Health and Human Services. No. 325.


Journal Abstracts

The following abstracts are reprinted verbatim as they appear in the cited source.

Cha, K.-W. & Weagley, R.O. (2002).Higher education borrowing.Financial Counseling and Planning,13(1), 61-73.

Using the 1992-93 Baccalaureate andBeyond Longitudinal Study and 1997follow-up, this study examined whichfactors influence the decision to borrowand the amount of borrowing for highereducation. A double-hurdle model wasapplied to analyze both the decisionto borrow and the borrowed amountequations. Current income and assetholdings had generally negative impactson higher education debt, whileexpected future income increasedamounts borrowed. Total costs had apositive effect on the probability andthe level of borrowing. Total grantsreceived had a negative influence onamount borrowed, but a positiveinfluence on the participation decision.

Cook, C.C., Crull, S.R., Fletcher,C.N., Hinnant-Bernard, T., &Peterson, J. (2002). Meeting familyhousing needs: Experiences of ruralwomen in the midst of welfarereform. Journal of Family andEconomic Issues, 23(3), 285-316.

Though sometimes overlooked, theavailability, affordability, and qualityof housing in rural communities are apotential barrier to transitioning fromwelfare to work. In this investigationwe examine housing issues confronting17 rural women and their families whowere recipients of welfare benefits in1997. Respondents’ housing accountsillustrate the significance of reliance onboth government housing subsidies andinformal subsidies supplied by friends,

family, and more distant relatives. Thestudy focuses on concerns women havein meeting their families shelter needsand the complexities involved in doingso. The findings of the research suggestthat additional housing policy initia-tives, as well as a targeted researchagenda are needed, especially forfamilies whose welfare benefits arenearing termination.

Lauderdale, D.S. & Kestenbaum, B.(2002). Mortality rates of elderlyAsian American populations basedon Medicare and Social Securitydata. Demography, 39(3), 529-540.

We present sex- and age-specificdeath probabilities for the elderly of sixAsian American subgroups—Chinese,Filipino, Indian, Japanese, Korean,and Vietnamese—based on data fromSocial Security Administration files.We determined ethnicity by combiningrace, place of birth, surname, and givenname. The data source and ethnicdetermination are the same for deathsand the population at risk, avoiding theproblem of noncomparability presentwhen data for the numerator come fromvital records and data for the denomina-tor come from census records. Wefound that death rates for elderly AsianAmericans are lower than those forwhites, and that socioeconomicdifferences between subgroups donot translate into like differences inmortality.

Lee, J. (2002). The poor in thefinancial market: Changes in theuse of financial products, institutions,and services from 1995 to 1998.Journal of Consumer Policy, 25(2),203-231.

Brewer, M.S. & Prestat, C.J. (2002).Consumer attitudes toward foodsafety issues. Journal of Food Safety,22(2), 67-83.

The objectives of this study were tosurvey consumer attitudes about thesafety of the food supply in general,relate general concern levels withgroups of specific items of concern,regulatory issues and prioritization offood safety funding areas, and tocompare these results with results of asimilar survey conducted in this lab in1994. Factor analysis of 360 consumerresponses showed six factors under-lying the 31 specific items evaluatedon individual 5-point scales (1 = nonconcern, 5 = very strong concern).MANOVA using general level of foodsafety concern (independent variable)were significant. Univariate testsshowed that as general level of concernwith food safety increased, so didconcern with chemical issues (artificialcolors, pesticide residues, hormones,preservatives, irradiated foods, exces-sive processing of foods, and plasticpackaging), spoilage issues (restaurantsanitation, shelf-stable foods, pasteur-ized foods, refrigerated, preparedfoods, improper food preparation,microbiological contamination andnutritional imbalances), health issues(vitamin, calorie, carbohydrate, fat,cholesterol and sugar content), regula-tory issues (pesticide safety, fish andimported food inspection, and healthlabeling of food), deceptive practices(naturally occurring toxins, foodingredients associated with allergiesand weight reduction diets advertisedas healthy) and information issues(availability of detailed informationat stores, markets and restaurants).

2002 Vol. 14 No. 2 99

The poor are in a disadvantagedposition in the financial market. In thisarticle, a review is given of publicpolicy initiatives that are implementedto help the poor as well as an examina-tion of how the poor are served in thefinancial market, using data from the1995 and 1998 Survey of ConsumerFinances provided by the FederalReserve Board. Specifically, poorhouseholds’ use of depository andcredit products, the financial institu-tions that provide these products to thepoor, and the way in which the poorconduct their financial business (e.g.,visit to branch offices, ATMs, etc.) arecompared to that of non-poor house-holds. Marketing and public policyimplications are drawn from thefindings.

Miyazaki, A.D. & Krishnamurthy, S.(2002). Internet seals of approval:Effects on online privacy policies andconsumer perceptions. The Journalof Consumer Affairs, 36(1), 28-49.

The use of Internet seal of approvalprograms has been touted as analternative to potential legislationconcerning consumer-related onlineprivacy practices. Questions have beenraised, however, regarding the effec-tiveness of such programs with respectto maintaining privacy standards andaiding online consumers. The authorsexamine these issues in a series of threestudies, the first of which is an explor-atory application of Federal TradeCommission privacy standards tovarious online privacy policies in aneffort to determine the ability of seal ofapproval program participation to act asa valid cue to a firm’s stated privacypractices. The second and third studiesare experiments designed to ascertainhow online firm participation inInternet seal of approval programsaffects consumers. Implications forconsumer policy are discussed.

Paeratakul, S., York-Crowe, E.E.,Williamson, D.A., Ryan, D.H., &Bray, G.A. (2002). Americans ondiet: Results from the 1994-1996Continuing Survey of Food Intakesby Individuals. Journal of theAmerican Dietetic Association, 102(9),1247-1251.

Objective. To examine the prevalenceof dieting to lose weight or for a healthreason in a representative sample of USadults.Design. Cross-sectional study design.Subjects/Setting. Data from 10,144participants of the 1994-1996 Continu-ing Survey of Food Intakes by Individu-als (CSFII 1994-1996) were used in theanalyses. All data were self-reported.Statistical analysis. Analysis included:cross-tabulation of dieting status bysociodemographic characteristics;comparison of the type of diet, thereason for dieting, and the source ofdiet used by men and women; compari-son of the nutrient intake and healthstatus of dieters and nondieters.Results. Prevalence of dieting variedby gender and race, being highest inwhite women (21%) and lowest inHispanic men (8%). About 71% of alldieters reported that they were dietingto improve health, and 50% reportedthat they were dieting to lose weight.Dieters reported lower intakes of totalfat, saturated fat, cholesterol, sodium,monounsaturated fat, polyunsaturatedfat, calcium, and selenium comparedwith nondieters. The rate of chronichealth conditions was higher amongdieters than nondieters. Self-reportedphysical activity was similar in bothgroups.Conclusions. The prevalence of dietingvaries according to sociodemographiccharacteristics. The reason for dietingand the type of diet used by dieters alsovary and need to be studied further. Ourresults suggest that the dieters generallyconsumed a more nutrient-dense dietthan the nondieters but still low incertain nutrients.

Nord, M., Andrews, M., & Winicki,J. (2002). Frequency and duration offood insecurity and hunger in UShouseholds. Journal of NutritionEducation, 34(4), 194-201.

Objective: This study examines theextent to which food insecurity andhunger in US households are occa-sional, recurring, or frequent/chronic.Design/Variables: The federal foodsecurity scale measures the severity offood insecurity in surveyed householdsand classifies households as to theirfood security status during the previousyear. The Current Population SurveyFood Security Supplement (CPS-FSS)collects the data elements used tocalculate the food security scale.Supplementary data on the frequencyof occurrence of the behaviors andexperiences comprising the foodsecurity scale are also collected by theCPS-FSS, but most of this informationis not included in the food securityscale. This study analyzes thesesupplementary data along with the foodsecurity scale and its constituent itemsusing data from the Food SecuritySupplement of the nationally represen-tative CPS conducted in August 1998.Results: About two thirds of house-holds classified as food insecure by thefederal food security scale experiencethe condition as recurring, and aroundone fifth experience these conditionsas frequent or chronic. The monthlyprevalence of hunger is about 60% ofthe annual prevalence, and the dailyprevalence is about 13% to 18% of theannual prevalence.Conclusions and Implications:Nutritionists can use these findings toenhance the informative value of foodinsecurity and hunger statistics fromnational, state, and local surveys wheninterpreting them to policy makers andto the general public.


Official USDA Food Plans: Cost of Food at Home at Four Levels,U.S. Average, December 2002 1

WEEKLY COST MONTHLY COST

AGE-GENDER Thrifty Low-cost Moderate- Liberal Thrifty Low-cost Moderate- Liberal GROUPS plan plan cost plan plan plan plan cost plan plan

INDIVIDUALS2

CHILD:1 year $16.60 $20.50 $24.10 $29.30 $71.90 $88.80 $104.40 $127.002 years 16.60 20.50 24.10 29.30 71.90 88.80 104.40 127.003-5 years 18.10 22.50 27.90 33.40 78.40 97.50 120.90 144.706-8 years 22.60 30.00 37.30 43.40 97.90 130.00 161.60 188.109-11 years 26.50 34.00 43.40 50.30 114.80 147.30 188.10 217.90

MALE:12-14 years 27.50 38.40 47.50 55.90 119.20 166.40 205.80 242.2015-19 years 28.50 39.60 49.30 56.90 123.50 171.60 213.60 246.5020-50 years 30.50 39.50 49.10 59.50 132.20 171.20 212.80 257.8051 years and over 27.70 37.60 46.20 55.50 120.00 162.90 200.20 240.50

FEMALE:12-19 years 27.50 33.10 40.10 48.40 119.20 143.40 173.80 209.7020-50 years 27.50 34.50 42.00 53.90 119.20 149.50 182.00 233.5051 years and over 27.10 33.60 41.70 49.70 117.40 145.60 180.70 215.40

FAMILIES: FAMILY OF 23:20-50 years 63.80 81.40 100.20 124.70 276.50 352.80 434.30 540.4051 years and over 60.30 78.30 96.70 115.70 261.10 339.40 419.00 501.50

FAMILY OF 4:Couple, 20-50 years and children—2 and 3-5 years 92.70 117.00 143.10 176.10 401.70 507.00 620.10 763.006-8 and 9-11 years 107.10 138.00 171.80 207.10 464.10 598.00 744.50 897.30

1Basis is that all meals and snacks are purchased at stores and prepared at home. For specific foods and quantities of foods in the ThriftyFood Plan, see Thrifty Food Plan, 1999, Executive Summary, CNPP-7A; for specific foods and quantities of foods in the Low-Cost,Moderate-Cost, and Liberal Plans, see Family Economics Review, No. 2 (1983). The Thrifty Food Plan is based on 1989-91 data andthe other three food plans are based on 1977-78 data; all four plans are updated to current dollars using the Consumer Price Index forspecific food items.2The costs given are for individuals in 4-person families. For individuals in other size families, the following adjustments are suggested:1-person—add 20 percent; 2-person—add 10 percent; 3-person—add 5 percent; 5- or 6-person—subtract 5 percent; 7- (or more)person—subtract 10 percent.3Ten percent added for family size adjustment.

2002 Vol. 14 No. 2 101

Consumer PricesAverage percent change for major budgetary components

Annual average percent change from Percent change December of previous year to December: 12 months ending

Group 1990 1995 2000 with December 2002

All Items 6.1 2.5 3.4 2.4Food 5.3 2.1 2.8 1.5 Food at home 5.8 2.0 3.0 0.8 Food away from home 4.5 2.2 2.4 2.3Housing 4.5 3.0 4.3 2.4Apparel 5.1 0.1 −1.9 −1.8Transportation 10.4 1.5 4.3 3.8Medical care 9.6 3.9 4.2 5.0Recreation NA 2.8 1.4 1.1Education and communication NA 4.0 1.2 2.2Other goods and services 7.6 4.3 4.5 3.3

Price per pound for selected food items

Price per pound unless otherwise noted (as of December in each year) DecemberFood 1990 1995 2000 2002

Flour, white, all purpose $ .24 $ .24 $ .28 $ .29Rice, white, long grain, uncooked .49 .55 NA .46Spaghetti and macaroni .85 .88 .88 .93Bread, white .70 .84 .99 1.03Beef, ground, uncooked 1.63 1.40 1.63 1.69Pork chops, center cut, bone-in 3.32 3.29 3.46 3.32Chicken, fresh, whole .86 .94 1.08 1.05Tuna, light, chunk 2.11 2.00 1.92 2.03Eggs, grade A, large, per dozen 1.00 1.16 .96 1.18Milk, fresh, lowfat, per gallon NA 2.31 2.66 NAButter, salted, grade AA, stick 1.92 1.73 2.80 2.84Apples, red delicious .77 .83 .82 .99Bananas .43 .45 .49 .50Oranges, navel .56 .64 .62 .74Potatoes, white .32 .38 .35 .48Lettuce, iceberg .58 .61 .85 .68Tomatoes, field grown .86 1.51 1.57 1.66Broccoli NA .76 1.52 1.05Carrots, short trimmed and topped .43 .53 NA NAOnions, dry yellow NA .41 NA NAOrange juice, frozen concentrate per 16 oz. 2.02 1.57 1.88 1.81Sugar, white, 33-80 oz. pkg. .40 .39 .40 .41Margarine, stick .87 .79 NA NAPeanut butter, creamy 2.09 1.78 1.96 1.90Coffee, 100% ground roast 2.94 3.75 3.21 2.84

NA = Data not available.Selected items from CPI Detailed Reports, Bureau of Labor Statistics, various issues. Price changes are for all urban consumers. Food prices areU.S. city average.

Consumer PricesConsumer Prices


Poverty Thresholds in 2000, by size of family and number of related children under 18 years

Related children under age 18 Eight

Size of family unit None One Two Three Four Five Six Seven or more

One person Under age 65 $9,359 Age 65 and over 8,628

Two people Householder under age 65 12,047 $12,400 Householder age 65 and over 10,874 12,353

Three people 14,072 14,480 $14,494Four people 18,556 18,859 18,244 $18,307Five people 22,377 22,703 22,007 21,469 $21,141Six people 25,738 25,840 25,307 24,797 24,038 $23,588Seven people 29,615 29,799 29,162 28,718 27,890 26,924 $25,865Eight people 33,121 33,414 32,812 32,285 31,538 30,589 29,601 $29,350Nine people or more 39,843 40,036 39,504 39,057 38,323 37,313 36,399 36,173 $34,780

Source: U.S Census Bureau, February 2003.

U.S. Poverty Thresholds and Related Statistics

Poverty Thresholds in 2002, by size of family and number of related children under age 18

9.6%

22.9%

27.1%

10.7%

Under 18 years

18 to 64 years

65 years and over

0 5 10 15 20

16.3%

10.1%

10.1%

Poverty rate by age, 2001

Source: U.S Census Bureau.

2002 Vol. 14 No. 2 103

Family Economics and Nutrition Reviewgratefully acknowledges the reviewers of manuscripts of 2002 issues.

P. Peter BasiotisCenter for Nutrition Policy and PromotionU.S. Department of Agriculture

Shanthy BowmanAgricultural Research ServiceU.S. Department of Agriculture

Nancy CanoltyUniversity of Georgia

Oral Capps, Jr.Texas A&M University

Sharon CristofarFood and Nutrition ServiceU.S. Department of Agriculture

Edward FrongilloCornell University

Sandria GodwinTennessee State University

David HaytowitzAgricultural Research ServiceU.S. Department of Agriculture

David HerringCenter for Nutrition Policy and PromotionU.S. Department of Agriculture

Helen H. JensenIowa State University

Rachel JohnsonUniversity of Vermont

Linda Scott KantorEconomic Research ServiceU.S. Department of Agriculture

Mark LinoCenter for Nutrition Policy and PromotionU.S. Department of Agriculture

Kristin MarcoeCenter for Nutrition Policy and PromotionU.S. Department of Agriculture

Margaret McDowellCenters for Disease ControlU.S. Department of Health and Human Services

Patricia McKinneyFood and Nutrition ServiceU.S. Department of Agriculture

Suzanne MurphyUniversity of Hawaii

Rodolfo NaygaTexas A&M University

Mark NordEconomic Research ServiceU.S. Department of Agriculture

Geoffrey D. PaulinBureau of Labor StatisticsU.S. Department of Commerce

Judith PutnamEconomic Research ServiceU.S. Department of Agriculture

Debra ReedTexas A&M University

Anita SinghFood and Nutrition ServiceU.S. Department of Agriculture

Leslie Speller-HendersonTennessee State University

Jamie StangUniversity of Minnesota

Guidelines for Submissions toFamily Economics and Nutrition Review

Family Economics and Nutrition Review (FENR) is a peer-reviewed journal published by the Center for Nutrition Policyand Promotion, United States Department of Agriculture. FENR will consider manuscripts concerning economic andnutritional issues related to the health and well-being of families. We are especially interested in studies about U.S.population groups at risk—from either an economic or a nutritional perspective. Research may be based on primary orsecondary data as long as it is national or regional in scope or of national policy interest. Subject matter should be basedon research findings of interest to a wide family economics and nutrition audience, including Federal, State, and localgovernment officials, nutrition and economic educators, and social scientists.

Your submission should contain the following:♦ an affiliation page that lists the author’s(s’) full name, academic degree(s), employer, and title. This list of names

must consist only of those who had an instrumental role in developing the manuscript being submitted.♦ a short abstract (about 15 lines) that summarizes the major findings. Abstracts are required for research articles,

not for research briefs.♦ text of 12 to 20 double-spaced pages for research articles or 5 to 10 double-spaced pages for research briefs.

Tables are single spaced. Articles over 20 pages in length will be considered by FENR editorial staff only inexceptional circumstances. Page limits include references but exclude author’s(s’) affiliation page, abstract page,tables, and graphs.

♦ no more than a total of five tables and graphs for research articles and two for research briefs to illustrate majorfindings. Do not include tables or graphs that are not referenced in the text. Tables larger than 1 full page will notbe considered. Tables and graphs labeled “1a, 1b, 1c,” for example, will count as three submissions.

♦ acknowledgment of the source of funding for the research.

Style:The writing style must be more journalistic than that used in purely academic journals. We encourage authors toreport descriptive statistics rather than multivariate analyses. We also encourage authors to use the active voice, toavoid jargon, to keep acronyms to a minimum, and to explain any technical terms. To be considered for publication,all manuscripts must follow the guidelines of the Publication Manual of the American Psychological Association,5th edition.

Format:FENR articles follow this general format: (1) abstract (for research articles only), (2) introduction, (3) methods, (4) results,(5) conclusions, (6) acknowledgments, and (7) references.

Tables, graphs, and other graphics should include titles in bold and sources at the bottom (if the data are from anothersource). Tables should be arranged to fit vertically (portrait style) on the page and should be done in a word processingprogram (Word, WordPerfect) by using tabs rather than a table function.

Refer to the Publication Manual of the American Psychological Association, 5th edition, for information concerning thecitation of references.

The font size of the text must be no smaller than 11 points; for tables, 10 points. Margins should be no smaller than1 inch.

Review:Research articles and briefs are peer-reviewed by a minimum of 2 professionals, with significant knowledge of the field,and are reviewed and edited by the FENR editorial staff.

Send your electronic copy to the managing editor: For specific questions or further information, [email protected] contact the editor or managing editor:

Phone: (703) 305-2732 or (703) 605-4435Fax: (703) 305-3300email: [email protected]

[email protected]

CENTER FOR NUTRITION POLICY AND PROMOTION Family Economics ... · PDF file2 Family Economics...

Documents

Transcript of CENTER FOR NUTRITION POLICY AND PROMOTION Family Economics ... · PDF file2 Family Economics...