A Longitudinal Study of Television Avertising Effects - Citir

9
A Longitudinal Study of Television Advertising Effects GEORGE P. MOSCHIS ROY L. MOORE* While many research questions regarding the effects of television advertising in consumer socialization require longitudinal research designs, nearly all previous research studies in the area have been cross-sectional or experimental. This article presents the results of a longitudinal study using a two-wave panel of adolescents with lag greater than a year. The study attempts to answer some questions regarding the effects of television advertising in the short run as well as in the long run. T he effects of television advertising, especially on chil- dren and adolescents, have come under fire in recent years from various public interest groups and researchers. Some advenising critics contend that advenising strongly influences the consumer attitudes, values, and behavior of youths and produces undesirable socialization such as non- rational choices, impulse-oriented decision-making, and the development of strong materialistic values. On the other hand, advenisers argue that the main sources of such be- haviors and cognitions are parents and peers, and that ad- venising may actually create positive consumer socializa- tion, such as socially desirable consumer behaviors and greater knowledge of the consumption process (cf. Ward 1979). To answer questions conceming the effects of television advenising on consumer behavior, researchers must seek three basic kinds of evidence (Seltiz, Wrightsman, and Cook 1959): Concomitant variation—correlation of television advenising with specific aspects of consumer be- havior Time order of occurrence—advenising' must occur before consumer behavior changes Elimination of other possible causal factors—elim- ination of other explanations of consumer behavior besides advenising Previous cross-sectional studies on television advenising effects on consumer socialization have satisfied only the •George P. Moschis is Associate Professor of Marketing, College of Business Administration, and Roy L, Moore is Assistant Professor of Joumatism, both at Georgia State University, University Plaza. Atlanta. GA 30303, This study was supported by research grants from the College of Business Administration and the Urban Life Center, Georgia State University, first condition, although cross-sectional studies do not allow for assessing directionality of the influence (e,g,, Roben- son, Rossiter, and Gleason 1979). Time order of occurrence has been addressed in several experimental studies (e.g., Goldberg and Gorn 1979; Goldberg, Gorn, and Gibson 1978). Aside from the drawbacks associated with research in laboratories (e.g., Murray 1980), the permanence of ad- venising effects has not been assessed in these studies (Adler 1977). Finally, interpersonal processes have been suggested as an altemative explanation of advenising ef- fects, since they may mediate the impact of advenising (Robenson 1979). Cognitive development has been sug- gested as an altemative explanation of social leaming (ad- venising and interpersonal) effects. For example, changes in a child or adolescent's consumer behavior may be due either to cumulative exposure to ads with age, to leaming from significant others, or to maturation and experience (Adler 1977). The weaknesses of previous research are reflected in the conclusions of Adler's (1977) review of television adver- tising effects. According to the repon, the effects of TV advenising on youths and their families were not clear, and several questions should be addressed in future research: • What are the long-term effects of television adver- tising on the development of consumer skills and other social benefits? • Does TV advenising contribute to "effective" or "good" consumer behavior pattems? • Does TV advenising contribute to any long-range socialization effects? • Are there any differences between long-term and shon-term advenising effects? What stereotyped beliefs result from heavy exposure to television commercials? 279 C JOURNAL OF CONSUMER RESEARCH • Vol, 9 • t)ecemt>er 1982

description

television advertising

Transcript of A Longitudinal Study of Television Avertising Effects - Citir

Page 1: A Longitudinal Study of Television Avertising Effects - Citir

A Longitudinal Study of TelevisionAdvertising Effects

GEORGE P. MOSCHISROY L. MOORE*

While many research questions regarding the effects of television advertising inconsumer socialization require longitudinal research designs, nearly all previousresearch studies in the area have been cross-sectional or experimental. Thisarticle presents the results of a longitudinal study using a two-wave panel ofadolescents with lag greater than a year. The study attempts to answer somequestions regarding the effects of television advertising in the short run as well asin the long run.

T he effects of television advertising, especially on chil-dren and adolescents, have come under fire in recent

years from various public interest groups and researchers.Some advenising critics contend that advenising stronglyinfluences the consumer attitudes, values, and behavior ofyouths and produces undesirable socialization such as non-rational choices, impulse-oriented decision-making, and thedevelopment of strong materialistic values. On the otherhand, advenisers argue that the main sources of such be-haviors and cognitions are parents and peers, and that ad-venising may actually create positive consumer socializa-tion, such as socially desirable consumer behaviors andgreater knowledge of the consumption process (cf. Ward1979).

To answer questions conceming the effects of televisionadvenising on consumer behavior, researchers must seekthree basic kinds of evidence (Seltiz, Wrightsman, andCook 1959):

• Concomitant variation—correlation of televisionadvenising with specific aspects of consumer be-havior

• Time order of occurrence—advenising' must occurbefore consumer behavior changes

• Elimination of other possible causal factors—elim-ination of other explanations of consumer behaviorbesides advenising

Previous cross-sectional studies on television advenisingeffects on consumer socialization have satisfied only the

•George P. Moschis is Associate Professor of Marketing, College ofBusiness Administration, and Roy L, Moore is Assistant Professor ofJoumatism, both at Georgia State University, University Plaza. Atlanta.GA 30303, This study was supported by research grants from the Collegeof Business Administration and the Urban Life Center, Georgia StateUniversity,

first condition, although cross-sectional studies do not allowfor assessing directionality of the influence (e,g,, Roben-son, Rossiter, and Gleason 1979). Time order of occurrencehas been addressed in several experimental studies (e.g.,Goldberg and Gorn 1979; Goldberg, Gorn, and Gibson1978). Aside from the drawbacks associated with researchin laboratories (e.g., Murray 1980), the permanence of ad-venising effects has not been assessed in these studies(Adler 1977). Finally, interpersonal processes have beensuggested as an altemative explanation of advenising ef-fects, since they may mediate the impact of advenising(Robenson 1979). Cognitive development has been sug-gested as an altemative explanation of social leaming (ad-venising and interpersonal) effects. For example, changesin a child or adolescent's consumer behavior may be dueeither to cumulative exposure to ads with age, to leamingfrom significant others, or to maturation and experience(Adler 1977).

The weaknesses of previous research are reflected in theconclusions of Adler's (1977) review of television adver-tising effects. According to the repon, the effects of TVadvenising on youths and their families were not clear, andseveral questions should be addressed in future research:

• What are the long-term effects of television adver-tising on the development of consumer skills andother social benefits?

• Does TV advenising contribute to "effective" or"good" consumer behavior pattems?

• Does TV advenising contribute to any long-rangesocialization effects?

• Are there any differences between long-term andshon-term advenising effects?

• What stereotyped beliefs result from heavy exposureto television commercials?

279C JOURNAL OF CONSUMER RESEARCH • Vol, 9 • t)ecemt>er 1982

vali
Highlight
vali
Highlight
vali
Highlight
vali
Highlight
Page 2: A Longitudinal Study of Television Avertising Effects - Citir

280 THE JOURNAL OF CONSUMER RESEARCH

Several of these questions require longitudinal research,but few longitudinal studies have been conducted, excepton the formation and persistence of brand loyalty (e.g.,Amdt 1971; Fauman 1966; Guest 1955, 1964; MadisonAvenue 1980). Thus, although longitudinal research is oftenadvocated, it has rarely been used to study consumer so-cialization in general and advenising effects in panicular(e.g.. Ward 1979; McLeod 1974).

This study examines the shon-term and longer term ef-fects of television advenising on the development of spe-cific consumption-related orientations in four areas: con-sumer role perceptions, normative consumer activities,materialistic values, and sex-role perceptions.

BACKGROUND AND HYPOTHESES

A conceptual model of consumer socialization developedin previous research (Churchill and Moschis 1979; Moschisand Churchill 1978; Moschis and Moore 1978, 1979) in-corporates five different types of variables: leaming prop-enies, age or life cycle, social structural variables, social-ization agents, and learning processes (Moschis andChurchill 1978). The five are classified as "antecedent vari-ables," "socialization processes," or "outcomes."

Antecedent variables include social structural and devel-opmental variables that locate the individual in his or hersocial environment. Examples of social structural variablesare social class, race, sex, and education; developmentalvariables include either age or life cycle. Socialization pro-cesses refer to agent-leamer relationships, which incorpo-rate the specific agent and leaming process. Socializationagents include mass media, parents, peers, and school,while learning processes include modeling (imitation ofleamer's behavior), reinforcement (positive or negative),and social interaction (which may include both modelingand reinforcement).

Outcomes in the model include the development of con-sumer knowledge, attitudes, and norms. Such orientationscan be categorized into (1) propenies that help the personfunction in any given social system and that are sociallydesirable and (2) propenies that are related to the individ-ual's behavior regardless of the social demands, includingsocially undesirable orientations (McLeod and O'Keefe1972).

Relationships between variables of the model are basedon socialization theories and previous research. Two of themost popular theories of socialization are cognitive devel-opment theory and social leaming theory. The cognitivedevelopment approach essentially views leaming as a cog-nitive-psychological process of adjustment to one's envi-ronment, with age used as a proxy variable for cognitivedevelopment. The social leaming model, on the other hand,focuses on sources of influence ("socialization agents")which transmit attitudes, motivations, and values to theleamer. Leaming is assumed to be taking place during theperson's interaction with socialization agents in various so-cial settings.

In the context of the social leaming approach to social-

ization, two theories of mass media influence seem to pre-vail in the literature that are relevant to the effects of tele-vision advenising on consumer socialization. One modelviews mass communication effects as powerful, with ex-posure to the mass media being "persuasive" per se. Thismodel makes the tacit assumption that media content pro-duces direct audience effects. Bandura (1971) argues thatthis rather simple stimulus-response model explains howmaterial objects acquire social meaning through mass mediaadvenising:

As a rule, observed rewards increase, and observed punish-ments decrease imitative behavior. This principle is widelyapplied in advertising appeals. In positive appeals, followingthe recommended action results in a host of rewarding out-comes. Smoking a brand of cigarettes or using a panicularhair lotion wins the loving admiration of voluptuous belles,enhances job performance, masculinizes one's self-concept,tranquilizes irritable nerves, invites social recognition andamicable responsiveness from total strangers.

According to the "limited effects" model (Bauer 1964;Klapper 1960), on the other hand, mass media reinforceexisting predispositions through selective exposure, andmass media effects are largely neutralized by interpersonalprocesses in a two-step flow. While the reinforcement view-has been attacked on several counts by socialization re-searchers (cf. Chaffee, Ward, and Tipton 1970), the "two-step-flow" process seems to be more directly related toconsumer socialization. Research findings suggest that themass media may induce youngsters to discuss consumptionmatters among themselves or with their parents and peers(e.g., Churchill and Moschis 1979; Moore and Stephens1975; Ward and Wackman 1971). Such mediation is morelikely to result in attitude formation and change than inreinforcement of existing attitudes (Chaffee et al. 1970;Ward and Wackman 1971).

If, as critics argue, television advenising has direct ef-fects on the consumer behavior of youths, then consumersocialization may occur—in line with the stimulus-responsemodel of mass media effects—regardless of mediating in-terpersonal processes. In this case, TV advenising exposureis expected to be related to the dependent measures in boththe shon run and the long run. However, it is possible thatexisting levels of cognitions may induce differential levelsof exposure/attention to TV advenising—i.e., reverse di-rection of influence in the shon run, as speculated in severalresearch studies (e.g., Adler 1977; Moschis and Churchill1978; Robenson et al. 1979), resulting in additional long-term leaming of such cognitions. This point is in a similarvein to that of research on television viewing and aggres-sion.' If this is the case, the panial correlations betweenTV advenising viewing at Time 1 and the dependent mea-

'According to this view, two most likely rival hypotheses are: (a) earlypreference for watching violent television content contributes to the (later)development of aggressive behavior; and (b) early aggression causes bothearly preference for violence viewing and later aggression (MurTay 1980.pp, 33-34),

Page 3: A Longitudinal Study of Television Avertising Effects - Citir

TELEVISION ADVERTISING EFFECTS 281

sures at Time 2 would be zero, after controlling for depen-dent measures at Time 1.

On the other hand, if consumer socialization takes placein line witli the "limited effects" model, then interpersonalprocesses should mediate advertising effects and the impactof television should be neutralized. In this case, the impactof advertising would be expected to vary by level of inter-personal communication. In addition, interpersonal pro-cesses would be expected to have a strong effect on con-sumer socialization, regardless of the level of TVadvertising viewing, in line with the "two-step-flow"model.

In summary, the main focus of this study is on short-term and longer term effects of television advertising onthe development of selected consumer orientations. Age,race, sex, and social class were used as antecedent (control)variables,^ and interpersonal communication about con-sumption with family and peers served as possible media-tors of television advertising effects. The four consumptionorientations related to adolescent consumer socialization ofinterests to various groups examined were: "positive" con-sumer role perceptions and consumer activities, which areviewed as socially desirable properties—i.e., as producingsocial benefits (Moore and Moschis, forthcoming; Moschisand Moore 1978, 1979); and materialism and sex-role ste-reotyping, which are seen as less desirable (e.g., Churchilland Moschis 1979; McLeod and O'Keefe 1972).

METHODS

SampleA two-wave panel study of adolescents with a 14-month

lag provided an opportunity to explore relationships be-tween television advertising and consumer leaming. Ado-lescents from several cities and towns in five counties inurban, suburban, semi-rural, and rural Georgia in juniorand senior high schools were asked to participate in a lon-gitudinal study by completing anonymous self-administeredquestionnaires. Specific schools were selected after per-sonal interviews with school officials to ascertain schoolsdemographically representative of their repsective regions.

Questionnaires were administered to 683 eligible respon-dents' in sixth through twelfth grades; a second wave ofquestionnaires was administered to a subsample of 230 ofthe original students approximately 14 months later. Severalof the students in the first wave were not included in thesecond wave due to graduation, absence, or relocation.Matching of the questionnaires was done using the respon-dent's birthdate and other demographics, whenever neces-

^Previous research showed that adolescence is an important period ofconsumer socialization (e ,g , , Moschis and Churchill 1978; Ward andWackman 1971), The selection of the antecedent variables was based uponprevious research showing that these background characteristics may affectthe way a young person responds to advenising ( e g , . Adler 1977: Chris-tiansen 1979; Moschis 1981),

'Student eligibility was based upon school policies regarding the use ofstudents and student information in surveys.

sary. The sample was generally representative with respectto sex (44 percent males and 56 percent females), age (59percent middle schoolers and 41 percent high schoolers),race (14 percent blacks and 86 percent whites), and socio-economic status measured on Duncan's scale (1961; mean= 50.3). These demographic characteristics are not verydifferent from the characteristics of samples used in pre-vious studies of consumer socialization (e.g., Churchill andMoschis 1979; Moore and Stephens 1975).* Because someof the questionnaires were incomplete or had errors in birth-days, the final usable sample consisted of 211 respondents.'

Definition and Measurement of Variables

Consumer role perceptions refer to the accuracy of theindividual's cognitions and perceptions of the consumerrole in terms of functions, obligations, (>osition, and rightsinvolved in role description (e.g., Moschis and Moore1978). Respondents were asked to indicate how much theywould or would not perform 11 behaviors associtated with(un)wise purchase and consumption of goods when theystart work and raise a family (e.g., "Check warranties andguarantees before buying" and "Buy throwaway bottlesinstead of retumable ones"). Responses were scored on afive-point "definitely would do" to "definitely wouldn'tdo" scale. High scores represent positive or socially desir-able consumer behavior. The alpha reliability coefficientswere 0.66 and 0.51 for Time 1 and Time 2 measures, re-spectively.

Consumer activity refers to the ability to buy and useproducts and services in a rational and efficient way (e.g.,Moschis and Churchill 1978). It was measured by summingresponses to 10 items on a five-point "very often" to"never" scale. Typical items were "I plan how to spend

'Although these demographic characteristics are fairly representative ofyouths in their respective regions, differences between those who com-pleted both surveys ("respondents") and those who completed only one("nomespondents") were still possible in the way they are affected bysocialization processes, especially advenising. To determine whether thetwo groups were affected differently by these prxxresses. the equality ofthe three regression coefficients (television viewing, family communica-tion, and peer communication) was tested for each criterion variable usingChow's (1960) f.test. The only significant difference that emerged waswith respect to the effects of socialization processes on sex-roles (F =4.07, p < 0,01). with family having a significant negative impact andpeers having a positive impact among "nonrespondents," The f-value formaterialism was 0,64 (n,s,). while the f-values for consumer role per-ceptions and consumer activity were 1,14 (n,s,) and 0,07 (n,s,). respec-tively. These results generally suggest no major differences between thetwo groups in the way they may be affected by these socialization pro-cesses in general and by television advenising in panicular,

'The numbers for the age categories filling out questionnaires at Tl andT2 were:

12 years old, n = 2313 years old, n = 4414 years old. n = 4115 years old. n = 4116 years old. n = 2017 years old, n = 3318 years old, n = 9

vali
Highlight
Page 4: A Longitudinal Study of Television Avertising Effects - Citir

282 THE JOURNAL OF CONSUMER RESEARCH

my money," "I carefully read most of the things they writeon packages or labels," and "I compare prices and brandsbefore buying something that costs a lot of money." Theindex could range from 10 to 50; its reliability coefficients,as measured by coefficient alpha, were 0.84 and 0.76.

Materialism is operationally defined as an orientationemphasizing possessions and money for personal happinessand social progress (Ward and Wackman 1971). This vari-able was measured by responses to six items such as "Itis really true that money can buy happiness." A five-point"strongly agree-strongly disagree" Likert-type scale wasused. The reliability coefficients of the scales were 0.71and 0.53.

Sex-role conceptions in family decision making refer tothe adolescent's perceptions regarding the relative influenceof husband and wife in family decisions. Perceptions weremeasured across 12 different decisions, representing dif-ferent degrees of husband-wife specialization. UsingHerbst's (1952) typology, two decisions were selected torepresent each one of six possible categories: Husband'sHousehold Duties (HH), Wife's Household Duties (HW),Common Household Duties (HB), Child Control and Care(Ch), Economic Activities (E), and Social Activities (S).

The following question was asked of respondents: "Ina family with small children, check who you think shouldhave the most say in deciding about the following things."The 12 decisions had four response altematives: "Husbandshould have most say," "Wife should have most say,""Husband and wife should have equal say," and "I don'tknow." The adolescent's perceptions of egalitarian sex-roles in family decision making were measured by summingresponses to the "Husband and wife should have equalsay" category to form a O-to-12 point index. The reliabilitycoefficients for the two measures were 0.65 and 0.72.

Because measures of television advertising based on theamount of time a person spends with the medium do notenable the researcher to separate programming from adver-tising effects, "closer" measures of the individual's fre-quency of interaction with socialization agents in generalhave been recommended by several researchers (e.g.,McLeod 1974; Ward et al. 1977). While product-specifictelevision advertising exposure measures are possible toconstruct (e.g., Robertson et al. 1979), measures of theindividual's total exposure to advertisements may well in-corporate several dimensions of the person's communica-tion behavior, such as frequency of exposure and motiva-tions for viewing (e.g., Atkin and Heald 1976).

In this research, one question was used to tap both ad-vertising viewing frequency and motivations for viewing.Thus, television advertising viewing was a direct measureof the adolescent's frequency of viewing TV commercialsfor the motives of gathering information for consumer de-cision making as well as information about life styles andbehaviors associated with consumer products. Respondentswere asked to indicate not only whether they watch tele-vision ads for various reasons (motivations) but also howoften they watch ads for such reasons (frequency). Theywere asked to indicate on a four-point "very often-never"

scale the extent to which they watched television ads forseven reasons, such as "to find out how good a product is"and "to find out what things to buy to impress others."*This measure of frequency along with motivations for in-teraction with the medium has been suggested by previoussocialization researchers (e.g., McLeod 1974) as a bettermeasure of television advertising than gross measures of"time spent with" or "frequency of viewing" television.Responses were summed across the seven items to form a7-to-28 point index, which had a reliability coefficient of0.83 for the Time 1 measure. Extemal validation of thismeasure was performed by correlating it with televisionviewing frequency as measured in previous studies (e.g.,Churchill and Moschis 1979). The correlation was 0.23(/? < 0.001).

Family communication about consumption was opera-tionally defined as overt interaction between parent and ad-olescent conceming goods and services (e.g., Moschis andChurchill 1978; Ward and Wackman 1971). It was mea-sured by summing responses to six items. A typical itemwas "My parents and I talk about buying things," withresponses measured on a five-point "very often" (5) to"never" (1) scale. The reliability coefficient alpha was0.62. "High" and " low" frequency groups were con-structed by splitting the scale at the median.

Peer communication about consumption was operation-ally defined as overt peer-adolescent interaction conceminggoods and services (e.g., Moschis and Churchill 1978:Moschis and Moore 1978). lt was measured by summingresponses to eight items such as "My friends and 1 talkabout buying things," on a five-point, "very often" (5) to"never" (1) scale. The reliability coefficent alpha was0.80. "High" and " low" frequency groups were con-structed by splitting the scale at the median.

RESULTSTable 1 shows product-moment and partial correlations

between television advertising viewing and the four depen-dent measures, both in the short run and the long run.Television advertising viewing correlates with role percep-tions and materialism in the short run. However, as was thecase in previous cross-sectional studies (e.g., Robertson etal. 1979), the direction of the influence is difficult to es-tablish. More valuable information may be obtained by ex-amining longer term television advertising effects—i.e..correlations between television advertising viewing at Time1 and the dependent variables measured at Time 2, espe-cially after controlling for Time 1 measures of the criterionvariables.

TV advertising viewing is not associated with consumerrole perceptions. This relationship remains insignificantafter controlling for previous levels of consumer role per-

The other items were: "to help me decide what things to buy," "tofind out where 1 can buy some things I want," "to have something to talkabout with others," "to leam about the Mn' things to buy," and "I seepeople on TV ads who are examples of the way I wish I were,"

vali
Highlight
vali
Highlight
vali
Highlight
vali
Highlight
vali
Highlight
Page 5: A Longitudinal Study of Television Avertising Effects - Citir

TELEVISION ADVERTISING EFFECTS 283

TABLE 1

TELEVISION ADVERTISING VIEWING AT TIME 1 (T1) ANDDEPENDENT MEASURES AT TIME 1 (Tl) AND TIME 2 (T2)

Product-moment Partialcorrelations correlations'

TABLE 2

TELEVISION ADVERTISING VIEWING AT TIME 1AND DEPENDENT MEASURES AT TIME 1 AND TIME 2

AMONG YOUNGER AND OLDER ADOLESCENTS

Consumer role perceptionsTlT2

Consumer activityT lT2

MaterialismTlT2

Sex-rolesTlT2

.35"

.05

.04

.07

,32".19'

,02-.10

-.08

.06

.09

-.12"

*Controllins lor measurement ol dependent vanable at Tl'p < 0,05'•p < 0,0;'p < 0.001

ceptions at Time 1 (r = -0.08). Similarly, the relationshipbetween TV advertising viewing and consumer activity inthe longer run is not significant (r = 0.07), and remainsinsignificant after controlling for previous levels of activitymeasured at Time 1.

Although the product-moment correlation between TVadvertising viewing and materialism is statistically signifi-cant (r = 0.19, p < 0.001), the correlation becomes in-significant when previous levels of materialism are par-tialled out. The correlation between egalitarian sex-roleperceptions and television advertising viewing approachessignificance (r = -0.10, p < 0.07); the correlation be-comes significant after controlling for previous levels ofthis measure at Time 1.

Because of possible age-related differences in televisionadvertising effects (e.g.. Ward and Wackman 1971), thedata were also analyzed by age groups. Table 2 showsrelationships between television advertising viewing anddependent measures among middle-school and high-schoolstudents. Tests of difference between correlations of thetwo age groups led to the acceptance of the null hypothesis,providing no support for the notion that younger adolescentsmay respond to television advertising differently than dotheir older counterparts.

These results suggest that television advertising may havefew direct effects in the long run. However, early exposureto television advertisements may be associated with laterdevelopment of materialism and traditional sex-roles, de-pending upon previous levels of such predispositions.

The effect of previously leamed cognitions on furtherleaming from television advertising was assessed by split-ting Time 1 (Tl) measures of the dependent variables into"high" and "low" levels of leamed cognition. It was ex-pected that, if existing cognitions lead one to pay attention

Consumer roleperceptions

TlT2

Consumer activityTlT2

MaterialismTlT2

Sex-rolesT lT2

Product-momentcorrelations

Younger

.13

.14

,40".16"

,38",07

- .08- .10

Older

- .12- .10

.25'

.2r

.35"

.07

- ,05- ,03

Partialcorrelations*

Younger

,08

,03

- ,05

- .12

Older

- .04

,20"

- .09

- .08

*Controlling lor measurement ol dependent vanable at Tl"p < 0,05'p < 0 01'p < 0.001

to television ads and leam consumer skills from them, thosewho scored high on these measures at Tl would be moreresponsive to television advertising in the long run.

Table 3 shows long-term relationships between televisionadvertising viewing measures at Tl and the criterion vari-ables at T2, by level of previously held cognitions andbehaviors at Tl . The data suggest that, among those whoscored low on consumer activity at Tl , television advenis-ing had significantly more negative effects on consumeractivity at T2 than among those scoring high at Tl (r =-0.22). The relationship remains unchanged after control-ling for other variables. Television advenising effects onmaterialism and sex-roles also appear to be the strongestamong those respondents who initially scored low on thesemeasures. The correlation between television advenisingviewing at Tl and materialism at T2 was 0.24, after con-trolling for antecedent variables, while the partial correla-tion between television advenising viewing at Tl and egal-itarian sex-role perceptions at T2 was -0.19. These datasuggest that the limited-effects communication model thatstresses perceptual defenses and reinforcement may not ap-ply to consumer socialization, highlighting the significanceof television advenising in the development and change ofconsumer orientations. The data also suggest that TV ad-vertising effects on consumer leaming may be contingentupon previous levels of leaming, with very limited selectiveperception operating.

To address the research question conceming the roles ofpeers and family as mediators of television advenising ef-fects, we analyzed the influence of television advenisingby level of communication with parents and peers. Table

Page 6: A Longitudinal Study of Television Avertising Effects - Citir

284 THE JOURNAL OF CONSUMER RESEARCH

TABLE 3

LONG-TERM RELATIONSHIPS BETWEEN TELEVISIONADVERTISING MEASURES (T1) AND DEPENDENT VARIABLES

(T2) BY PREVIOUS LEVEL OF DEPENDENT MEASURE (T1)

Control variableProduct-moment

correlationPartial

correlation*

Role perceptions (T1)Low(n = 110)High(n = 101)

Consumer activity (Tl)Low (n = 99)High(n = 112)

Materialism (Tl)Low(n = 113)High (n = 98)

Sex-roles (Tl)Low (n = 121)High (n = 90)

.01-.06

.06

.23-

.12

-.20'.02

.02- .10

-.22=- .02

.24'

.09

-.19"-.05

*Partialing out the eflects ol age, race, sex, social dass and communication with parentsand peers

"p < 0,0S'p < 0,01

4 shows long-term and short-term relationships betweentelevision advertising and the dependent variables by levelof interpersonal communication.

The relationship between television advertising and con-sumer role perceptions is strong both among families inwhich interpersonal communication about consumption isfrequent (r = 0.20) and among those in which it is infre-quent (r = 0.37) only in the short run, but the direction ofthe influence is not clear. Apparently, long-term televisionadvertising effects on role perceptions are independent ofthe level of family communication. The relationship be-tween TV advertising and consumer activity is not signif-icant in either the long run or the short run among bothtypes of families. The relationship between TV advertisingand materialism is significant in the short run, both amongfamilies in which discussion of consumption matters is in-frequent (r = 0.17) and among those in which it is frequent(r = 0.35). However, in the long run, the influence oftelevision advertising is statistically significant only amongfamilies in which communication about consumption occursinfrequently (r = 0.22). These data suggest (hat parentsmediate some television advertising effects by discussingconsumption matters with the child.

On the role of peers as mediators of advertising effects,the data showed strong positive relationships in the shortmn between TV advertising viewing and role perceptionsand materiaJism, regardless of the level of peer interaction.Few long-term effects emerged, suggesting selective ex-posure to communication. Thus, the "limited effects"model of mass communications may apply only in casesinvolving the leaming of less desirable consumer orienta-tions, such as materialism and traditional sex-role stereo-types, and only in families in which consumption mattersare discussed infrequently.

TABLE 4

SHORT-TERM AND LONG-TERM TELEVISION ADVERTISINGEFFECTS BY LEVELS OF INTERPERSONAL COMMUNICATION

Interpersonal communication

Family (T1)* Peers (Tl)"

Low High Low High(n = 108) (n = 103) (n = 92) (n = 119)

Role perceptionsShort-term (Tl)Long-term (T2)

Consumer activityShort-term (Tl)Long-term (T2)

MaterialismShort-term (Tl)Long-term (T2)

Sex-rolesShort-term (T1)Long-term (T2)

.20=

.02

.00

.08

!22°

- .01-,18=

,15

,05.03

.35*

.14

.11- .08

.41*- .11

.16

.00

.24°

.13

.07- .09

.20°- ,05

.11

.06

.35'

.04

.03- .13

"Entries are partial correlations, controlling lor age, sex, race, social class, peer com,munication an6 previous level of respective dependent variable at Time 1 (T1) in analyzinglong-term effects.

'Entries are partial correlations, controlling for age, sex, race, social dass. family com,munication and previous level of respective dependent variable at T1 in analyzing long-term effects.

'p < 0.05'p < O.Ot•p < 0.001

Finally, to examine the extent to which interpersonal in-fluences affect consumer socialization in isolation from TVadvertising and to find out whether advertising serves as acatalyst in the process (Adler 1977, p. 131), we analy-sayzed the impact of interpersonal communication pro-cesses on the dependent measures by levels of TV adver-tising viewing. The results of these analyses are shown inTable 5.

The effects of family interaction on consumer role per-ceptions are significant only in the short run and only amongadolescents frequently viewing TV ads (r = 0.21). Familyinteraction effects are also likely to increase the adoles-cent's propensity to perform socially acceptable consumerbehaviors in the short run in the presence of heavy TV adviewing (r = O.I 9); Jong-run effects of famiJy on consumeractivity are more likely to occur in the absence of TV ad-vertising viewing (r = 0.18).

The family is likely to have some influence on the ado-lescent's development of egalitarian sex-roles in the shortrun to the extent that there is little viewing of TV advertis-ing (r = 0.16), but long-term effects are more likely tooccur in the presence of frequent TV advertising exposure.

Peer communication effects also appear to be weaklycorrelated with the level of TV advertising exposure.Among those adolescents with low television advertisingviewing, peer communication is positively related to thedevelopment of consumer role perceptions (r = 0.20) andmaterialism (r = 0.17). Among those adolescents with fre-

Page 7: A Longitudinal Study of Television Avertising Effects - Citir

TELEVISION ADVERTISING EFFECTS 285

TABLE 5

SHORT-TERM AND LONG-TERM INTERPERSONAL COMMUNICATION RELATIONSHIPSWITH DEPENDENT MEASURES BY LEVEL OF TV ADVERTISING VIEWING*

Role perceptionsShort-term (Tl)Long-temn {T2)

Consumer activityShort-term (T1)Long-term (T2)

MaterialismShort-term (Tl)Long-term (T2)

Sex-rolesShort-temi (T1)Long-term (T2)

Family communication (Tl)

TV ad viewing

Low (n = 107)

.08

.03

.12

.18"

- .10.07

.16"- .05

frequency (Tl)

High (n = 104)

.21"

.05

.19'- .07

.01- .07

- ,07,20"

Peer communication (Tl)

TV ad viewing

Low (n = 107)

.20"

.00

.09

.12

.17"- .08

- .06.08

frequency (TI)

High (n = 104)

.11

.16"

.02

.24'

.08

.03

.02-.19"

•Entries are partial correlations, controlling lor the effects of age. sex, raoe, social dass, peer communication or family communication, and previous level of leaming in analyzing long.fermeffects,

"p < 0,05'p < 0,01

quent exposure to TV ads, the effects of interpersonal dis-cussion appear to be long-lasting, affecting their consumerrole perceptions (r = 0.16), consumer activity (r = 0.24),and the development of traditional sex-role perceptions indecisiot\ making (r = -0.19).

DISCUSSIONSome short-run effects of TV advertising emerged in this

study, but since the direction of the influence is not clearand the correlations not particularly strong, these resultsshould be interpreted with caution. Some long-term tele-vision advertising effects also emerged, but in many casesleaming from television was associated with previous levelsof such leaming. Specifically, increased advertising expo-sure seems to decrease the person's likelihood of perform-ing socially desirable consumer behaviors, but only amongthose adolescents who are not likely to perform such activ-ities in the first place. It does seem to contribute to thedevelopment of materialistic values and traditional sex-roleperceptions among those who have not yet developed suchpredispositions. These findings are not consistent with theselective exposure/perception hypothesis advanced in stud-ies of television violence and aggression (Murray 1980),but are consistent with findings in the related area of polit-ical socialization (Atkin and Gantz 1978).

On the question of whether television advertising has adirect effect or is mediated through interpersonal processes,it was found that the family communication environmentmay play a sigrvificant role. Specifically, television adver-tising appears to affect the development of materialism and

traditional sex-roles when parents do not discuss consump-tion matters with their children, perhaps placing the ado-lescent child at the mercy of advertising, a finding consis-tent with previous research (e.g., Churchill and Moschis1979; Comstock 1978).

Television advertising effects on adolescents interactingwith peers were observed in the short run, but these resultscast some doubt on the direction of such influence. Appar-ently, peers play a minor mediating role in leaming frommass communication, at least in the long run.

Whether family and peer interaction result in effectiveleaming appears to be conditioned to some extent by thefrequency of the adolescent's viewing of television com-mercials. Thus, leaming from television may be a second-order consequence of interpersonal communications, withtelevision serving as a catalyst. This result appears to beconsistent with longitudinal findings in political socializa-tion (Atkin and Gantz 1978): it is also in line with specu-lations about leaming from television based on cross-sec-tional data (Moschis and Churchill 1978).

In summary, the data suggest that television advertisingmay have both short-term and long-term effects on con-sumer socialization. Some negative effects may occur di-rectly, especially among families lacking interpersonalcommunication about consumption. Families discussingconsumption are likely to neutralize such effects. The me-diating role of television ads appears to be complex. Thedata suggest that interpersonal processes may condition ayouth's attention to and leaming from television commer-cials, resulting in both negative and positive socialization.Finally, the data suggest that the effects of television ad-

vali
Highlight
vali
Highlight
vali
Highlight
Page 8: A Longitudinal Study of Television Avertising Effects - Citir

286 THE JOURNAL OF CONSUMER RESEARCH

vertising may be different in the short run than in the longrun, suggesting the need for separating short-term fromlong-term effects.

Because only a few dependent variables were used in thisstudy, the findings may not apply to consumer socializationin general. Furthermore, other explanatory variables orthird variables may account for the findings, even thoughthe explanatory variables included in the present study havebeen suggested by theory and previous research. While thisstudy does not provide answers to all questions regardingtelevision advertising effects on consumer socialization, itaddresses such questions and uses longitudinal rather thancross-sectional data. Through the use of longitudinal studiessuch as this one, we can begin answering some of the press-ing questions conceming television advertising effects.

[Received August 1981. Revised July 1982.]

REFERENCESAdler, Richard P, (1977), "TV Effects on Consumer Skills and

Interpersonal Processes," Washington, DC: U.S. Govem-mctit Printing Office.

Ardnt, Johan (1971). "A Research Note on IntergenerationalOverlap of Selected Consumer Variables." Markeds Kom-munikazion. 3, 1-8.

Atkin, Charles K. and Walter Gantz (1978), "Television Newsand Political Socialization," Public Opinion Quarterly.42(2), 183-198.

and Gerald Heald (1976), "Effects of Political Advertis-ing," Public Opinion Quarterly. 40, 216-228.

Bandura. Alben(1971), "Modeling Influences on Children." tes-timony before the Federal Trade Commission (November),cited in: Scott Ward (1974). "Consumer Socialization,"Journal of Consumer Research. 1 (September), p. 7.

Bauer, Raymond A. (1964), "The Obstinate Audience: The In-fluence Process from the Point of View of Social Commu-nications," American Psychologist. 19 (May). 319-328.

Chaffee, Steven H,, L, Scott Ward, and Leonard P, Tipton(1970). "Mass Communication and Political Socialization,"Journalism Quarterly. 47, 647-659.

Chow, Gregory C. (1960), "Tests of Equality Between Sets ofCoefficients in Two Linear Regressions." Econometrica. 28(July), 591-605.

Christiansen, ;ohn B. (1979), "Television Role Models and Ad-olescent Occupational Goals" Hurnan Communication Re-search. 5 (Summer), 335-338,

Churchill, Gilbert A., Jr. and George P. Moschis (1979), "Tele-vision and Inteqxrsonal Influences on Adolescent ConsumerLeaming," Journal of Consumer Research. 6 (June), 23-35.

Comstock, George (1978), "The Impact of Television on Amer-ican Institutions," Journal of Communication. 28 (Spring),J2-48.

Duncan, Otis D. (1961), "A Socioeconomic Index of All Occu-pations," in Occupations and Social Status, ed. Albert J.Reiss, Jr., New York: The Free Press.

Fauman, B. C. (1966), "Determinants of Adolescents' BrandPreferences," unpublished dissertation, Sloan School ofManagement, MIT, Cambridge, MA.

Goldberg, Marvin E. and Gerald J. Gom (1979), "TelevisionImpact on Preferences for Non-White Playmates: Canadian'Sesame Street' Inserts," Journal of Broadcasting. 23 (Win-ter), 27-32.

, Gerald J, Gom, and Wendy Gibson, (1978), "TV Mes-sages for Snack and Breakfast Foods: Do They InfluenceChildren's Preferences?" Journal of Consumer Research. 5(September), 73-81.

Guest, Lester P, (1955), "Brand Loyalty: Twelve Years Later,"Journal of Applied Psychology. 39 (December), 405-408.

(1964), "Brand Loyalty Revisited: A Twenty-Year Re-port," Journat of Applied Psychology. 48 (April), 93-97.

Herbst, P. G. (1952), "The Measurement of Family Relation-ships," Human Relations. 5 (February), 3-25.

Klapper, Joseph T. (I960), The Effects of Mass Communication.New York: Free Press.

Madison Avenue (1980), "Seventeen Makes a Sales Call," 22(November), p. 85, 88-91, 94-96.

McLeod, Jack M. (1974), "Commentaries on Ward, 'ConsumerSocialization,"'Journal of Consumer Research. 1 (Septem-ber), 15-16.

and Garret J. O'Keefe (1972), "The Socialization Per-spective and Communication Behavior," in Current Per-spectives in Mass Communication Research, eds. F. G. Klineand P. J. Tichenor, Beveriy Hills, CA: Sage Publications,121-168.

Moore, Roy L. and George P. Moschis (forthcoming), "The Roleof the Mass Media and the Family in the Development ofConsumption Norms," Journalism Quarterly.

and Lowndes F. Stephens (1975), "Some Communicationand Demographic Determinants of Adolescent ConsumerLeaming," Journal of Consumer Research. 2 (September),80-92.

Moschis, George P. (1981), "Socialization Perspectives and Con-sumer Behavior," m Review of Marketing 1981. eds. B, Enisand K. Roering, Chicago: American Marketing Association.43-56.

and Gilbert A. Churchill. Jr, (1978), "Consumer Social-ization: A Theoretical and Empirical Analysis," Journal ofMarketing Research. 15 (November). 599-609,

and Roy L. Moore (1978), "An Analysis of the Acqui-sition of Some Consumer Competencies Among Adoles-cents," Journal of Consumer Affairs. 12 (Winter). 277-279,

and Roy L. Moore (1979). "Decision Making Among theYoung: A Socialization Perspective," Journal of ConsumerResearch. 6 (September), 101-112.

Murray, John P. (1980), Television & Youth: 25 Years of Researchand Controversy. Boys Town, NB: The Boys Town Centerfor Study of Youth Etevelopment.

Robenson, Thomas S. (1979), "Parental Mediation of TelevisionAdvenising Effects," Journal of Communication. 29 (Win-ter), 12-25.

, John R, Rossiter, and Terry C. Gleason (1979), "Chil-dren's Receptivity to Proprietary Medicine Advenising,"Journal of Consumer Research. 6 (December), 247-255.

Seltiz, Claire, Lawrence S, Wrightsman, and Stuan W, Cook(1959), Research Methods in Social Relations (revised edi-tion). New York: Holt, Rinehan & Winston.

Ward, Scott (1979), "Researchers Look at the 'Kid Vid' Rule:Overview of Session," in Advances in Consumer Research.Vol. 6, ed. W. L. Wilkie, Ann Arbor, MI: Association forConsumer Research, 7-11.

and Daniel Wackman (1971), "Family and Media Influ-ence on Adolescent Consumer Leaming," American Behav-ioral Scientist. 14 (January-February), 415-427.

, Daniel B. Wackman, and E. Wanella (1977), How Chil-dren Learn to Buy: The Development of Consumer Infor-mation Processing Skills. Beverly Hills, CA: Sage.

Page 9: A Longitudinal Study of Television Avertising Effects - Citir