Articulo Pr on Tua Rio

CULTURAL MODELS OF DOMESTIC VIOLENCE: PERSPECTIVES OFSOCIAL WORK AND ANTHROPOLOGY STUDENTS

Cyleste C. CollinsCase Westem Reserve University '- ,

William W. DressierUniversity of Alabama

This study employed a unique theoretical approach and a series of participant-

based ethnographic intervievñng techniques that are traditionally used in cog-

nitive anthropology to examine and compare social work and anthropology

students' cultural models of the causes of domestic violence. The study findings

indicate that although social work students and anthropology students share

understandings of a general model of domestic violence, social work students

agree on distinctive elements of this model that anthropology students do not.

These findings are important in better understanding the role sodal work edu-

cation plays in developing sodal workers' understandings of the roots of

domesfic violence.

DOMESTIC VIOLENCE IS A pervasive sodal prob-

lem that can demand attenfion from a variety

of sodal service providers (see Pyles, 2006;

Pyles & Postmus, 2004; Tjaden & Thoennes,

2000). Sodal workers who come into contact

with families who are affected by domestic

violence can provide these families with hope

by linking them with important services.

Some research has found, however, that social

workers hold biases and stereotypes about

domestic violence (Danis & Lockhart, 2003;

Ross & Glisson, 1991), and that they frequent-

ly fail to provide necessary services to victims

(Eisikovits & Buchbinder, 1996; Kok, 2001).

Research has suggested that it is important to

identify workers' ideas about the causes of

and appropriate treatment for domestic vio-

lence in tackling these issues (Davis, 1984;

Davis & Carlson, 1981). The aim of this arficle

is to examine the cultural models of domesfic

violence shared by sodal work students and

other social science students and to introduce

methods with which to eiidt and examine

those cultural models.

Even when victims of domesfic violence

do not seek shelter or other services from do-

mestic violence agencies, they often become

engaged with the social welfare system in a

Journal of Social Work Education, Voi. 44, No. 2 (Spring/Summer 2008).Copyright <̂ 2008, Council on Social Work Education, Inc. All rights r«serv«d. 53

54 JOURNAL OF SOCIAL WORK EDUCATION

number of other ways, including when theyneed financial and other types of assistance(Brandwein, 1999; Raphael, 2001), or are facedwith questions about the welfare of their chil-dren (Edleson, 1999; Kohl, Edleson, English, &Barth, 2005; Postmus & Ortega, 2005).Although recent progress in social policy hasincreased funding for and awareness ofdomestic violence, research continues to indi-cate that victims of domestic violenceencounter a number of barriers in their inter-actions with hun\an service professionals.

Research has found that even whendomestic violence victims disclose their abusestatus to their caseworkers, they often feeluncomfortable doing so (Busch & Wolfer,2002; Saunders, Holter, Pahl, Tolman, &Kenna, 2005). Some welfare workers inappro-priately screen (Levin, 2001; Postmus, 2000;Postmus, 2004; Owens-Manly, 1999) or other-wise fail to identify clients eligible for access-ing the Family Violence Option (McNutt,Carlson, Rose, & Robinson, 2002) or similarprograms that allow domestic violence vic-tims to obtain waivers for time limits andother restrictions enacted under the PersonalResponsibility and Work Opportunity Recon-ciÜation Act of 1996 (P L. 104-193). Welfareworkers also tend to underestimate the num-bers of victims that apply for assistance (Kok,2001) and sometimes provide inappropriateservices when encountering victims (Brand-wein, 1999). Some welfare offices have ad-dressed this problem by placing domestic vio-lence advocates in welfare offices and trainingstaff to be aware of the barriers that victims ofdomestic violence face. These programs havemet with varying levels of success (Kok, 2001),but are steps in the right direction. As profes-

sionals become educated about domestic vio-lence, victims' challenges in accessing theservices to which they are entitled shoulddecrease.

Improving victims' access to servicesrequires better understanding how profes-sionals think about and approach domesticviolence cases. One way to investigate this isto ask whether particular groups of profes-sionals share ideas, ideas that might havebeen formulated on the basis of having under-gone common training and therefore result ina professional culture. Answering such ques-tions requires that we deconstruct the conceptof culture. Recent research has pointed out theproblematic nature of social work's usageof the term "culture" (see Park, 2005). Cul-ture has been commonly understood insocial work and other fields to be an all-encompassing term (D'Andrade, 1999), some-times standing in for racial and ethnic charac-teristics (e.g., Hispanic culture), as well asaspects of the environment (e.g., office cul-ture) (Park, 2005). Common to these descrip-tions of culture is the idea of sharing. For morethan 100 years, anthropologists have debatedthe definition of culture, and in recent years,have developed techniques for systematicallyassessing it. Many cognitive anthropologistshave come to agree that culture is best definedas shared knowledge among individuals in agroup (D'Andrade, 1984; Shore, 1996). Such adefinition allows for greater specificity in nar-rowing down, operationally defining, andmeasuring elements of culture. In this study,techruques traditionally employed in cogni-tive anthropology are used to identify the spe-cific ways social work students think aboutdomestic violence.

CULTURAL MODELS OF DOMESTIC VIOLENCE 55

Cuiturai Modeis and Cognitive

Anthropoiogy

Anthropologists have theorized that culturalknowledge consists of a collection of models,or a set of interlocking schémas that guideindividuals in interpreting and responding totheir environments (D'Andrade, 1984). Eachindividual's model is thought to be composedof not only individual biographical and idio-syncratic information, but also that which isculturally transmitted, or leamed, and there-fore shared with other members of the cultur-al group (Shore, 1996). Thus, although sharingis central to understanding culture, thisframework also accounts for individualityand thus, intracultural diversity (i.e., variabil-ity in shared cultural models within a socialgroup) (Pelto & Pelto, 1975).

To examine the structure of cultural mod-els, cognitive anthropologists often employtechniques that avoid making assumptionsabout the content of any given cultural model.Instead, emergent techniques are used to elic-it such content from a culture's participants.Some of the research methods drawn on toevoke such content are free listing, con-strained and unconstrained pile sorts, rank-ing, and rating tasks (Weiler & Romney, 1988).These techniques build on one another andallow researchers to gather progressively morestructured information about how participantsthink about the given domain of inquiry. Thefirst phase, "free listing," is an open-endedtechnique in which participants respond to theresearcher's prompt, generating a list of ideasabout the cultural domain in their own words.The second phase, "pile sorting," requires par-ticipants to organize their thinking systemati-

cally by creating categories of meaning. Thefinal stages of rating and ranking tasks allowthe researcher to adopt sophisticated analytictechniques to examine the data quantitatively,including cultural consensus analysis. Cul-tural consensus analysis evaluates the extentto which the data are considered cultural data,that is, shared, or are simply unique to anindividual. The aim of all these techniques isto describe cultural meaning in the terms themembers of a social group themselves use,and not to impose the understanding of theresearcher on that group.

Cuiturai Consensus Analysis

The cultural consensus model, conceptualizedby Romney, Weiler, and Batchelder (1986)assumes that the content of cultural modelscan be determined by systematically inter-viewing participants. Cultural consensusanalysis is a factor analytic-type techniquethat supplies three sets of results. First, itextracts the eigenvalues of a subject-by-subject correlation matrix; if the ratio of thefirst-to-the-second eigenvalue is sufficientlylarge (interpreted using a rule of thumb—thatthe ratio of the first factor to the second is atleast 3.0), this indicates that there is sufficientsharing among individuals to conclude thatthey are using the same cultural model.Second, each individual's cultural compe-tence is calculated, evaluating each person'slevel of sharing with the group model.Calculating a cultural competence coefficientenables the researcher to identify the "ex-perts" in the group (i.e., those whose know-ledge is closest to the group's aggregated or"average" knowledge). Finally, cultural con-sensus analysis reveals what is sometimes


referred to as the cultural answer key, that dis-

closes what the culturally agreed-on, or "cor-

rect" answers were, or what the group mem-

bers agreed were the components of the culture.

Consensus analysis is often performed

using ANTHROPAC software (Borgatti, 1988).

It is important to note that these analyses are

not subject to the usual sample size require-

ments and assumptions of statistical analysis.

In fact, many of these, such as the assumption

of independence of observations, are consid-

ered inappropriate for cultural data, in which

cultural participants are specifically assumed

to share beliefs. Tests of violations of such

assumptions have indicated that the use of

small sample sizes does not jeopardize the use

of statistics and subsequent interpretation of

cultural data (Handwerker, Hatcherson, &

Herbert, 1997).

Important insights about the connections

between culture, health, and health behavior

(Chavez, Hubbell, McMullin, Martinez, &

Mishra, 1995; Chavez, McMullin, Mishra, &

Hubbel, 2001; Dressier, Dos Santos, & Balieiro,

1996), culture and poverty, and culture and its

effects on organizations (Caulkins & Hyatt,

1999; Jaskyte & Dressier, 2004) have been

gained using the cultural consensus model.

Until now, however, the model has not been

applied to domestic violence research, and has

been rarely used in sodal work.

Purpose of the Study

This study sought to identify elements of stu-

dents' cultural models of domestic violence,

to examine the structure of these models, and

to evaluate the degree to which a cultural

model of domestic violence is shared among

them. Examining sodal work students while

they are undergoing training to become

human service professionals provided a useful

point of entry in understanding the extent to

which social work training shapes their beliefs,

as well as similarities and differences in how

human service professionals might define and

respond to domesfic violence. A group of

anthropology students served as a comparison

group, under the assumption that, as social sci-

ence students, anthropology students would

have some understanding of domestic vio-

lence, but that their understandings would not

be codified the way that social work students'

understandings would be, given their status as

service providers-in-training.

The study incorporated a framework for

examining attitudes and beliefe to determine

whether they constitute cultural models that has

been little used in sodal work research, but has

great potential for improving the understanding

and measurement of person-in-environment.

The concept of cultural models provides for a

more detailed and person-focused analysis of

domestic violence attitudes than has been previ-

ously attempted, and ultimately will provide

insight into the ways in which sodal work edu-

catioa policy and practice regarding domestic

violence can be improved.1

Methodi • '• •-

Consistent with cognitive anthropology meth-

odology, a series of standardized data collec-

tion techniques were used to examine partid-

pants' cultural models. These procedures are

described in detail elsewhere (Weiler & Rom-

ney, 1988), but focus on elidting partidpants'

beliefs about the causes of domestic violence.


Data were collected in three phases. Because

each stage of the data collection involved a

unique set of participants, techniques, and

analysis, each will be discussed as a separate

study. Only minimal descriptive information

was collected from participants until the third

phase of the research, when demographic

characteristics became important issues in the

analysis.

Phase 1: Eliciting Elements of theDomain

Participants and procedure. In the first phase of

the research, participants generated the terms

associated with domestic violence through

free listing. The sample consisted of 25

advanced-standing social work students en-

rolled in a graduate level research course, all

but one of whom was female. Participants

were asked to list as many causes of domestic

violence that they could think of, and write

them on a blank sheet of paper.

Analysis. Participants listed a total of 29

different causes of domestic violence (see

Table 1). One phrase, "poor education on ap-

propriate behavior" was dropped because of

its wordiness and lack of clarity, and a total of

28 terms were retained for analysis and use

in subsequent phases of the research. The 28

items were analyzed to determine which

terms participants mentioned most frequently,

and which were most salient (i.e., how early in

students' lists they appeared, indicating the

terms' primacy) (see Table 1). The analysis

indicated that financial difficulties, substance

abuse, having witnessed abuse, stress, and

power were both the five most often men-

tioned and most salient causes (those items

that are listed earliest), with more than 50% of

the participants listing these terms.

Phase 2: identifying the SemanticStructure of the Domain

Participants and procedure. In the second phase

of the study, a total of 40 students participated

in the research. Of these, 24 were graduate-

level social work students enrolled in a

research course, and 16 were upper-division

and graduate students enrolled in an anthro-

pology course. Using the 28 terms generated

from participants in the first phase of the

research, participants in the second phase com-

pleted an unconstrained pile sort to examine

how they organized their ideas about domestic

violence. Participants were told that the terms

they would sort were various potential causes

of domestic violence (which primed them to

categorize on this basis), and were told to place

terms they believed were similar into piles

together. Participants were told that they could

make as many or as few piles as they liked.

Once the terms were sorted, the researcher con-

ducted open-ended group interviews with the

participants, in which they were asked to give

a name or theme to each of their piles that

would explain why they had placed particular

terms together in these respective piles.

Analysis. Pile-sort data were submitted to

multidimensional scaling (MDS) techniques

to identify the possible dimensions partici-

pants used in classifying the domestic vio-

lence terms (Hair, Anderson, Tatham, & Black,

1996; Kruskal & Wish, 1978). MDS transforms

proximity data (i.e., ratings of similarity) to

extract a visual picture of participants' group-

ings of terms. Stress values in MDS assess the


TABLE 1. Free-List of Causes of Domestic Vioience as Reported by SociaiWoric Students {N=2S), by Frequency and Saiience

ItemMention

Frequency%

Mentioned Salience

Financial difficulfies

Substance abuse

Having witnessed abuse

Stress

Power

Anger

Having been abused

Low self-esteem

Work-related stress

Family-related stress

Jealousy

Lack of control

Mental illness

Unemployment

History of abuse

Low education level

Isola fi on

Poor communication

Infidelity

The culture of violence

Poor coping skills

Lack of respect for partner

Marital problems

Selfishness

Unstable living condifions

Unhappiness

Anxiety

Depression

1918171413107776655543333322211111

76

72

68

56

52

40

28

28

28

24

24

20

20

20

16

12

12

12

12

12

8

8

8

4

4

4

4

4

.539

.479

.398

.346

.298

.181

.120

.093

.182

.151

.176

.087

.084

.136

.094

.066

.054

.077

.059

.049

.021

.028

.071

.023

.007

.021

.010

.014

goodness of fit of a given dimensional solu-

fion; the higher the stress value, the poorer the

solufion's fit. In other words, a high stress

value (greater than 0.20) indicates that the

"mapped" MDS poorly represents the similar-

ities calculated in the original similarity

matrix. A two-dimensional solution yielded

stress values below 0.20 for the separate

analyses of social work and anthropology stu-

dents, as well as for their merged data.


Cluster analysis was performed next toidentify specific groupings within the MDSboundaries (Hair et al., 1996). Generally, par-ticipants tended to group together items thatwere relevant to individual issues (e.g.,depression, unhappiness, selfishness, sub-stance abuse, mental illness). Other groupsincluded personal history/experiences (in-cluding having been abused and having wit-nessed abuse), and issues external to the indi-vidual (e.g., money, work stress, low level ofeducation). The relationships between theterms in the MDS and cluster analysis are rep-resented in two din\ensions in Figure 1.

Phase 3: Examining Dimensions ofMeaning

Participants and procedure. A total of 51 students

participated in the rating/ranking phase of theresearch. Of these, 34 were master's-levelsocial work students enrolled in a researchcourse, and 17 were a mix of graduate andundergraduate students (11 graduates, 6undergraduate) ervolled in a graduate-levelanthropology course. The participants weremostly White (85.7%) and female (74.5%).Several significant differences were foundbetween the student groups with regard totheir demographic characteristics. More socialwork students (55.9%) reported havingworked full-time in a human service agency(compared to 0% of anthropology students;X'(l, «=51)^15.14, p<.001), social work stij-dents were significantly older (M=29.7,SD=9.1) than the anthropology students(M=23.2, SD=4.3), f(48)=2.80, p<.001), andbecause all social work students were gradu-ate students but only one third of anthro-

pology students were, there was also a signifi-cant difference between the participants on thebasis of class level. (x^(l, ii=50)= 29.97, p<.001).

Four dimensions of meaning were takenfrom the pile sorts and their interviews for usein the third phase of the study. First, because itis a basic evaluative dimension, participantswere asked to rank the importance of thecauses of domestic violence. Participantsranked the terms from most to least impor-tant, where the most important cause wasgiven a score of one, and the least important,a score of 28. Next, participants had indicatedin interviews and in their pile-sort data thatthey believed some terms went togetherbecause they were either "inside" or "outside"of the victim or perpetrator. These werereferred to in the third phase of the research ascauses that were "internal and external" to theindividual, and were evaluated on a 5-pointLikert-type scale, where l^internal, and 5=external Other participants had noted thatthey had placed items in the same pilebecause the cause was one that could bechanged. Thus, the third dimension waslabeled "amenability to change," and wasevaluated on a 5-point Likert-type scale onwhich l=not amenable to change, and 5=highlyamenable to change. Finally, because severalitems were mentioned that referred to mentalhealth, the fourth dimension participantsevaluated was "mental health issues." Thisdimension was evaluated using a simple'l=yes/0=no choice format.

Analysis. The data analysis was conduct-ed in two phases. First, cultural consensusanalyses were performed separately on eachstudent group to determine the extent to


FiGURE 1. Muitidimensionai Scaiing and Ciusters: Sociai Work andAnthropoiogy Students' Understandings of tlie Relationsiiips Among DomesticViolence Causes

1.00

0.50

ËQ

-0.00

-0.50

-1.00

mental illnessstress

unhappinessdepressionunemployment

money

education

unstable living conditions

anxiety

low self-esteem

isolation

poor coping skills

selfishness

jealousy

angercommunication problemsmarital problems

family problemspower

lack of control

substance abuse

lacks respect for partner

violent cultureabused

abuse historywitnessing abuse

having abused others

-1.00 -0.50 0.00

Dimension 1

0.50 1.00

Note. The axes of the figure refer to each dimension of the 2-dimensional configuration on which

the multidimensional scaling of the terms was based. Though mental health and substance abuse

were often placed in piles together according to cluster analysis, the terms are not well represent-

ed in space by a 2-dimensional multidimensional scaling analysis solution.


which each shared a model of domestic vio-lence causes, and then on the entire sample'sdata to determine whether all students shareda model. Findings from the separate consensusanalyses of the sodal work students revealedthat they did, in fact, share a model of the caus-es of domestic violence on all four dimensionsof meaning. Anthropology students, however,shared a model on only two dimensions:amenability to change and the importance ofthe domestic violence cause. The merged datarevealed a pattern that mirrored theanthropology-only analysis (see Table 2); sodalwork students and anthropology students

shared a model on the dimensions of amenabil-ity to change and importance, but not on theinternality/externality of the causes or whetherthey were mental health issues. On the inter-nal/external dimension and mental healthissues dimension, sodal work students' levelsof cultural competence were both less variableand higher overall than that of anthropologystudents. In fact, an examination of the errorbars on these two dimensions, sorted by group,demonstrated that on the internal/externaldimension, sodal work and anthropology stu-dents' competence levels overlapped very lit-tle, and on the mental health dimension, the

TABLE 2. Cultural Consensus Analysis for Social Work Students, AnthropologyStudents, and Full Sample: Levels of Agreement (Cultural Competence) on FourDimensions of Meaning

Dimension

Importance*Internal/external*Amenability to change*Mental health *

Importance*Internal / ExternalAmenability to changeMental health

Importance*Internal/ExternalAmenability to changeMental health

EigenvalueRatio

Cultural CompetenceAf

Social Work Students

5.022.894.54

4.53

Anthropology :

.4902.372.451.75

38.56

.55

.51

Students

.53

.43

.47

.46

Total Sample

4.952.534.232.70

.55

.48

.51

.45

SD

.20

.22

.24

.17

.37

.36

.28

.20

.26

.29

.26

.19

CompetenceRange

.17-.87-.05-.87.a9-.910-.85

-1.0-1.0-.50-.83

0-1.0

-.15-.78

-.42-.84

-.44-.83-.13-.91

0-.88

Note. Possible Competence Range - -1.0-1.0. Asterisk (*) indicates sufficiently highto conclude a shared model exists.

competence


two groups did not overlap at all. These find-ings further emphasize the distinctiveness ofthe two groups of students. Table 3 is a briefsummary of the content of the culturalmodels—only the five highest- and lowest-ranked and rated terms on each dimension arelisted except for mental health. The mentalhealth issue column lists examples of termsparticipants evaluated as being mental healthissues and ones that they did not.

Following the consensus analysis, PROF-IT (PROperty-FITting) analysis (Carroll &Chang, 1964; Kruskal & Wish, 1978) was per-formed to connect the consensus results withthe MDS representation of the similarities ofthe pile-sort terms, by determining whether

participants actually used the identifieddimensions in their pile-sort ratings of theterms' similarity. PROFIT analysis assumesthat data are distributed linearly and increasemonotonically (Borgatti, 1994), as in regres-sion analysis. The coordinates of the MDS arethe independent variables in a regressionanalysis performed in PROFIT, and the attrib-utes, derived from the consensus analysisanswer key, are the dependent variables(Borgatti, 1996). PROFIT analysis yields sever-al important pieces of information. First, themultiple R explains the overall measure of fitof a given attribute; a multiple R closer to 1.0indicates a better fit. Next, coordinates are giventhat represent the head of an arrow of a vector

TABLE 3. Content of Cuiturai Model: Most and Least Important Factors byDimensions of Meaning (Merged Data)

MostImportant

Abuse historyHaving been

abusedHaving abused

othersPower

Substance abuse

LeastImportant

Anxiety

DepressionSelfishness

UnhappinessLow education

Most/Internal

AngerLow self-esteem

JealousyAnxietyStress

Most/External

Work stressFinancial

difficultiesUnstable living

conditionsUnemploymentThe culture of

violence

Most Amenableto Change

Marital problemsUnemployment

Poor coping skillsDepression

Communicationproblems

Least Amenableto Change

Having been abusedHaving abused

othersHaving witnessed

abuseAbuse historyThe culture of

violence

Mental HealthIssue

Depression

AnxietyHaving been abused

UnhappinessHaving witnessed

abuse

No MentalHealth Issue

UnemploymentFamily problems

Financial difficultiesPower

Selfishness


that will run through the multidimensional

space. This vector evaluates how well the attrib-

utes represented the pile-sort data. In the social

work students' data the multiple K's for three of

the four dimensions of meaning were greater

than .75 {see Table 4). For anthropology stu-

dents, however, the multiple R's were signifi-

cant for only two dimensions: amenability to

change and importance, consistent with the

consensus analysis findings. Thus, the PROFIT

analysis suggests that sodal work students

made use of three of the four dimensions in

their pile sorts, but only the amenability to

change dimension and the importance ranking

dimensions were used by anthropology stu-

dents. Figure 2 displays the results of the PROF-

IT analysis on the MDS graph.

Dimension independence. The next step in

the analysis was performed to determine the

different dimensions' relationships to each

other. The answer keys were correlated to test

the extent to which the dimensions were inde-

pendent of one another. The correlations for

anthropology and social work students'

answer keys mirrored each other, and so will

be discussed in general terms. The internal/

external dimension was unrelated to both im-

portance and amenability to change, but was

moderately negatively related to mental

health (r=-.39 for social work, r=-.46 for an-

thropology). Meanwhile, mental health was

somewhat negatively related to amenability to

change (r--.26 for sodal work, r--.22 for

anthropology) and importance {r=-.22 for

TABLE 4. PROFIT Analysis: Understanding How Dimensions of Meaning Relateto Pile-Sort Data, by Student Type

Measure ofSignificance

Multiple R

R'

Multiple R

R'

Multiple R

R'

Multiple R

R^

Social Work Students Anthropology Students(«=34) (rt=17)

Importance

.79**

.63

Internal/External

.83**

.69

Amenability to Change

.76**

.57

Mental Health Issue

.30

.09

.73*

.53

.43

.18

.72**

.52

.15

.02

All Students(N=51)

.82**

.67

.78**

.60

.80**

.63

.35

.20

Note. *p<.01, **p<.001.


FIGURE 2. PROFIT Analyses: How Participants Used Dimensions of Meaning InOrganizing Their Beliefs About Domestic Violence.

1.00

0.50

o

1ï -0.00

-0.50

-1.00

mental illnessstress

work stress

unemployment

money

educationIMPORTANCE

unstable living conditions

communication problems

unhappinessdepressionanxiety

low self-esteem

isolation

selfishness

jealousy

marital problems

family problems

substance abuse

AMENABLE TO CHANGE

violent culture

INTERNAL / EXTERNAL anger

power

lack of control

lacks respect for partner

having been abi isecabuse history

witnessing abusehaving abused others

-1.00 -0.50 0.00

Dimension 1

0.50 1.00

Note. Causes of domestic violence located closest to the arrowhead were estimated to have a

higher degree of the given dimension of meaning.


social work, r=-.28 for anthropology), butthere was a moderately positive correlationbetween importance and amenability tochange {r=.52 for social work, r=.77 foranthropology). None of these correlations arestrong enough to threaten the independenceof the dimensions, but do point out someinteresting relationships among them. Spe-cifically, terms that were ranked as mostimportant were rated as less amenable tochange and were likely to be evaluated asmental health issues (recall that a ranking of 1in importance indicates high importance),and terms that were considered mental healthissues were rated as less amenable to change,and more as internal causes of domesticviolence.

Examining the distribution of sharing. The

next step in the analysis was to examine howsharing in the sample was distributed, that is,whether there were characteristics other thanstudent group that predicted sharing.Significant demographic characteristics of thesample—gender, age, year in school, and hav-

ing worked full-time in a human serviceagency—were entered into multivariate GLManalyses as covariates, using competence coef-ficients from each of the four merged data setsas the dependent variables, and student typeas the main independent variable (see Table5). Findings revealed a significant main effectof gender on the importance dimension; beingfemale predicted having significantly highercompetence, and a significant main effect ofstudent type for the internal/external dimen-sion; social work students had higher compe-tence. The analysis of the amenability tochange dimension revealed a significant maineffect of working in a human service agency;students who had not worked in a humanservice agency had significantly higher com-petence. On the mental health dimension, novariables exerted significant main effects, butstudent type was marginally significant(p=.O9), and reached tradifional levels of sig-nificance when nonsignificant variables weredropped from the model, with social workstudents demonstrating higher competence

TABLE 5. F-Tests, Multivariate General Linear Modei for Dimensions of Meaning

Variable Importance*Internal/External''

Amenabilityto Change^

Mental HealthIssue**

Student typeGender

Age

Graduate student

Full dme student in a humanservice agency

.0112.34***

2.43

.38

1.73

5.38**

.01

.02

.10

1.04

1.05

2.57

2.28

.01

4.58***

3.03*

.13

1.04

.19

1.37

Note. *p<.10, **p<.05, ***p<.001.


than anthropology students on this dimen-sion. While these findings are interesting andconsistent with previous analyses, the low R^values on each of the dimensions' models sug-gest there are factors that remain unaccountedfor, so the results should be interpreted withsome caution.

Discussion

This study was undertaken to examine andcompare cultural models of domestic violenceof social work and anthropology students.The research employed a new and person-focused framework to gain an understandingof how social work education prepares stu-dents to deal with domestic violence. Theñndings represent a significant step towardunderstanding the development and mainte-nance of human service professionals' beliefsabout domestic violence, with the ultimategoal of improving the delivery of services andeliminating barriers victims face. The data col-lection worked to uncover participants'beliefs about the causes of domestic violencebecause service providers' beliefs about thecause of a person's predicament tend to pre-dict their responses to that person (Corrigan &LWatson, 2003). Identifying such beliefs could,in turn, help explain some of the difficultiesdomestic violence victims report in dealingwith service providers.

A Cultural Model of Domestic

Violence

The evidence in this study is consistent withthe notion that anthropology and sodal workstudents share a cultural model of domesticviolence, but that social work students make

finer gradations among some elements of thatmodel. The findings indicated that on thedimensions of importance and amenability tochange, social work and anthropology stu-dents demonstrated strong agreement, pro-viding evidence for the existence of a folkmodel of domestic violence. Folk models are"tacit forms of knowledge" that are "moreconservative than scientific theories and aremore resistant (though not completely resist-ant) to empirical disconfirmation" (Shore,1996, p. 65), shared among individuals, andimphcit (D'Andrade, 1987). As noted earlier inthis article, simple sharing of particular ideasis only one piece of the cultural model puzzle;raüier, the extent of sharing and departuresfrom that sharing (intracultural diversity) arealso important. Specifically, in this study, twobackground characteristics, having experiencein a human service agency, and gender, moder-ated the extent to which the students sharedideas on amenability to change and importance.

That those participants who had notworked full-time in a human service agencyhad the highest levels of agreement on theamenability to change dimension explicitlypoints to the relevance of nonexpert knowl-edge (i.e., folk knowledge) in interpreting thepossibility of change in domestic violence sit-uations. It may be that those with experiencesin human service agencies recognize either agreater or lesser number of possibilities forintervention effectiveness. The finding thatwomen demonstrated more agreement on theimportance dimension is consistent with pre-vious domestic violence research that hasfound gender differences in attitudes (seeWorden & Carlson, 2005; Carlson & Worden,


2005), but should be interpreted with cautionsince few men participated in this study. Here,abuse history, having witnessed abuse, andhaving abused others were considered to beboth the most important causes of domesticviolence and also the causes least amenable tochange, suggesting that the students wereskeptical about the efficacy of domestic vio-lence interventions.

A Professionally Eiaborated Modei ofDomestic Vioience

While this study's findings indicated thatsocial work and anthropology students sharea folk model of domestic violence, they alsoindicated that social work students' under-standing was unique, based on their consen-sus on two dimensions: (1) the extent to whichthey thought causes of domestic violence areinternal or external to the individual, and (2)whether they believed those causes are mentalhealth issues. Social work students also usedthese dimensions in explaining their catego-rizations of domestic violence causes in a wayanthropology students did not. Such a uniqueimderstanding might be understood as socialwork students' making finer distinctions, orelaborating on elements of the folk model.

In the course of their education and train-ing, social work students become attuned toaspects of the interaction between a personand his or her environment, aspects that arelikely to include understandings of mentalhealth issues, and issues that are internal tothe individual (i.e., contributed by the per-son), as well as those that are consideredexternal but still part of the person's envi-ronment. Social work students' training is

undertaken in the context of the classroomand "book learning," as well as through learn-ing and practicing clinical skills. Thus, it islogical that sodal workers might have a morerefined understanding of the relevance ofmental health and the intemality / externalityof domestic violence issues. Human behaviorin the social environment, a core course insocial work curricula, focuses explicitly onfactors, both internal (in the individual) andexternal (in the environment) as causes ofbehavior. At the same time, many social workprograms offer course électives in psy-chopathology and mental health, offer con-centrations in mental health, and infusemental-health concepts into their generalistcoursework. With these considerations inmind, it might not be surprising to find thatsocial work students share ideas on particulardimensions of meaning.

It is tempting to conclude, based on thesefindings, that this research has revealed aunique professional cultural model of domes-tic violence among social-workers-in-training.A few issues must be considered before such aconclusion can be drawn, however. First, a cul-tural model is a complex structure, composedof a network of schematized representations.As such, capturing that model by measuringonly a few dimensions of an issue as multifac-eted as domestic violence is probably unrealis-tic. Second, the fact that all the student partici-pants shared understandings on particular ele-ments of the model indicates that social workand anthropology students' understandingswere not completely different. This might notbe surprising because both were students inthe sodal sciences and therefore could share


ideas on that basis. Third, because the sodalwork students had not yet received their pro-fessional degrees, it is unclear whether themodel applies to professionals in practice. Thesocial work students' understandings mightbe better described as elaborations of a folkmode! of domestic violence that could eitherbe the result of their professional trairûng, orthe folk model could be a simplified version ofa professional model. Future research on themodels employed by professional socialworkers would be well-positioned to clarifythis issue.

The existence of distinct professionalmodels of domestic violence is certainly plau-sible, given different professionals' disagree-ments about the causes and appropriate han-dling of domestic violence cases (Davis, 1987),even while the mechanisms by which such amodel exerts its influence remain unclear.Little research has been conducted on socialworkers' professional socialization (Barreta,2004), and thus how specific domestic vio-lence education and training are manifested inpractice is uncertain (Tower, 2003). It is impor-tant to note, too, that even if social work stu-dents share ideas about domestic violence as aresult of their social work education, suchsharing is not entirely unproblematic.Specifically, Danis and Lockhart (2003) havenoted that when social workers focus on indi-vidual and mental health issues as causes ofdomestic violence, they risk engaging invicfim-blaming behaviors and further trauma-tizing their clients. To avoid this, researchershave suggested that it might be useful tointegrate domestic violence education intopractice and core courses in the social work

curriculum {see Begun, 1999; O'Keefe & Men-

nen, 1998).

Strengths, Limitations, andDirections for Future Research

This study is one of the first to employ the cul-tural consensus model to understanding theeffects of social work educafion on students,and as such, works to improve the social workresearch knowledge base on the subject ofculture. The study is also the first of its kindto use the cultural consensus model in study-ing domesfic violence, and thus makes asignificant contribution to the literature inthe area. The cultural consensus model, aunique, theory-centered approach, is a naturalfit for social work because of its focus onthe person-in-envirorunent and use of open-ended participant-centered methodology.This study, consistent with the methodologi-cal techniques of cognitive anthropology, useda mixed-methods approach in achieving itsgoals, allowing stronger conclusions to bedrawn (Teddlie & Tashakkori, 2003).

While the findings of the research areprovocafive and suggest that social work edu-cafion has a unique impact on the beliefs of itsstudents, the study findings should be inter-preted cautiously. The comparison group ofanthropology students was small and hetero-geneous (their class levels were mixed), andthe social work and anthropology group sam-ple sizes were unequal. Before we can con-clude that social work uniquely influences itsstudents' cultural models, future researchshould invesfigate the models of studentsfrom a number of different fields, including avariety of professional programs, the sciences,


and liberal arts fields. Such research wouldalso lend more credence to the existence of afolk model of domestic violence. This studyshould also be replicated in other areas of thecountry and with larger sample sizes toimprove its generalizability.

Other limitations to this research shouldbe noted. First, anthropology students did notcomplete separate free lists of domestic vio-lence causes; instead, they worked from thosegenerated by the sodal work students. Thus,there is no way to know whether anthropolo-gy students were working simply within asocial work framework, or whether theywould have generated different terms alto-gether. Second, background characteristicswere not collected at every stage of theresearch. Although such data were not used inanalysis until the final phase (and thereforenot collected until then), they might have beenuseful to better understand the background ofthose completing the earlier stages of thestudy. Finally, information about students'exposure to domestic violence course contentor personal experiences with domestic vio-lence was not collected. Previous researchsuggests that this is an important considera-tion in fully understanding how people con-ceptualize and deal with domestic violence(see Collins, 2005; Goldblatt & Buchbinder,2003; Nabi & Homer, 2001), and should beincluded in future research.

An additional issue with these data con-cerns the perpetual chicken-and-egg problem;that is, it is unclear whether students whoenter social work programs are different uponprogram entry, or whether their sodal workeducation shapes their attitudes and beliefs.

Recent research has begun to uncover the spe-cial influence social work education has onstudents. One study examined students at thebeginning and the end of their social workeducation and found that sodal work educa-tion influences attitudes toward women(Black, 1994). Future research replicating thisstudy could sort out this dilemma by examin-ing students at different points in their sodalwork education; that is, measuring students'beliefs at both the beginning and the end oftheir professional education. Such a studywould provide insight into the contributionsodal work education makes to its students'professional development.

If future research finds unique culturalmodels among social work students, it will beimportant to identify the particular aspects ofsociai work education that fadlitate the for-mation and maintenance of such a model. Thedevelopmental trajectory of such models alsoneeds to be addressed, as well as the extent towhich they are amenable to change throughtraining. This issue is espedally important toprograms seeking to improve services todomestic violence victims and to remove bar-riers to service.

Conclusion

This study's use of a new method of studyingattitudes and beliefs provides significant hopefor better understanding the problems domes-tic violence victims encounter in their interac-tions with social service professionals.Specifically, idenfifying how professionals-in-training conceptualize and respond to domes-tic violence should help lead to educationalmodels that improve service delivery. Such


changes would assist vicfims of domesfic vio-

lence and their families in rebuilding their

lives, prevenfing social service systems and

the professionals who work in them from

being hindrances to progress.

References

Barretfi, M. (2004). What do we know about

professional socialization of our students?

Journal of Social Work Education, 41,

255-283.

Begun, A. L. (1999). Infimate partner violence:

An HBSE perspecfive. Journal of Social

Work Education, 35, 239-252.

Black, B. M. (1994). Students' atfitudes toward

women: Do social work programs make a

difference? Affilia, 9, 417-436.

Borgatü, S. (1988). ANTHROPAC: An MS/DOS

program with documentation. University of

California, Irvine.

Borgatti, S. P. (1994). Cultural domain analy-

sis. Journal of Quantitative Anthropology, 4,

261-278.

Borgatti, S. P. (1996). ANTHROPAC 4.0

Methods Guide. Natick, MA: Analyfic

Technologies.

Brandwein, R. (1999), Family violence and

welfare use: Report from the field. In R.

Brandwein (Ed.), Battered women, children,

and welfare reform: The ties that bind.

Thousand Oaks, CA: Sage.

Busch, N. B., & Wolfer, T (2002). Battered

women speak out: Welfare reform and

their decisions to disclose. Violence

Against Women, 8, 566-584.

Carlson, B. E., & Worden, A. P. (2005).

Attitudes and beliefs about domestic vio-

lence: Results of a public opinion survey.

I. Definifions of domesfic violence, crimi-

nal domesfic violence, and prevalence.

Journal of Interpersonal Violence, 20,

1197-1218.

Carroll, J. D., & Chang, J. (1964). A general

index of nonlinear correlation and its

applicafion to the interpretafion of mulfi-

dimensional scaling solutions. American

Psychologist, 19, 540.

Caulkins, D., & Hyatt, S. B. (1999). Using con-

sensus analysis to measure cultural diver-

sity in organizations and social move-

ments. Field Methods, 11(1), 5-26.

Chavez, L. R., Hubbell, F. A., McMullin, J. M.,

Marfinez, R. G., & Mishra, S. I. (1995).

Structure and meaning in models of

breast and cervical cancer risk factors: A

comparison of percepfions among Latin-

as, Anglo women, and physicians. Medical

Anthropology Quarterly, 9(1), 40-74.

Chavez, L. R., McMullin, J. M., Mishra, S. I. &

Hubbell, F. A. (2001). Beliefs matter:

Cultural beliefs and the use of cervical

cancer screening tests. American Anthro-

pologist, 303,1114-1129.

Collins, C. C. (2005). Cultural models of domestic

violence: Perspectives of human service pro-

fessionals. Doctoral dissertation. Univer-

sity of Alabama.

Corrigan, P.W., & Watson, A.L. (2003). Factors

that explain how policy makers distribute

mental health resources. Psychiatric

Services, 54, 501-507.

D'Andrade, R. G. (1984). Cultural meaning

systems. In R. A. Schweder and R. A.

Levine (Eds.), Culture theory: Essays on

mind, self, and emotion (pp. 88-Î19). Cam-

bridge: Cambridge University Press.


D'Andrade, R. G. (1987). A folk model of the

mind. In D. Holland & N. Quinn (Eds.),

Cultural models in language and thought

(pp. 112-148). Cambridge; Cambridge

University Press.

D'Andrade, R. G. (1999). Culture is not every-

thing. In E. L. Cerroni-Long (Ed.), Anthro-

pological theory in North America. Westport,

CT: Bergin & Garvey.

Danis, E, & Lockhart, L. L. (2003). Domesüc vio-

lence and sodal work education: What do

we know, what do we need to know? Jour-

nal of Social Work Education, 39, 215-224.

Davis, L. V. (1984). Beliefs of service providers

about abused women and abusing men.

Social Work, 29, 243-251.

Davis, L. V. (1987). Views of wife abuse: Does

the research method make a difference?

Affilia, 2(2), 53-66.

Davis, L. V., & Carlson, B. E. (1981). AtHtudes

of service providers toward domestic vio-

lence. Social Work Research and Abstracts,

26(4), 34-39.

Dressier, W. W., Dos Santos, J. E., & Balieiro,

M. C. (1996). Studying diversity and shar-

ing in culture: An example of lifestyle in

Brazil. Journal of Anthropological Research,

52, 331-353.

Edleson, J. L. (1999). The overlap between

child maltreatment and woman battering.

Violence Against Women, 5(2), 134-154.

Eisikovits, Z., & Buchbinder, E. (1996).

Pathways to disenchantment: Battered

women's views of their sodal workers.

Jourrml of Interpersonal Violence, 11,425-440.

Goldblatt, H-, & Buchbinder, E. (2003).

Challenging gender roles: The impact on

female social work students of working

with abused women. Journal of Social Work

Education, 39. 255-275.

Hair, J. F., Anderson, R. E., Tatham, R. L., &

Black, W. C. (1996). Multivariate data analy-

sis with readings. New York: MacMillan.

Handwerker, W. P., Hatcherson, J., & Herbert,

J. (1997). Sampling guidelines for cultural

data. CAM Journal: Cultural Anthropology

Methods, 9{n 7-9.

Jaskyte, K., & Dressier, W. W. (2004). Studying

culture as an integral aggregate variable:

Organizafional culture and innovation in

a group of nonprofit organizations. Field

Methods, 36(3), 265-284.

Kohl, P. L., Edleson, J. L., English, D. J., &

Barth, R.P. (2005). Domestic violence and

pathways into child welfare services:

Findings from the National Survey of

Child and Adolescent Well-Being. Children

and Youth Sennces Review, 27, 1167-1182.

Kok, A. C. (2001). Economic advocacy for sur-

vivors of domestic violence. Affilia, 16,

180-197.

Kruskal, J. B., & Wish, M. (1978). Multi-

dimensioiml scaling: Quantitative applica-

tions in the social sciences. Newbury Park,

CA: Sage.

Levin, R. (2001). Less than ideal: The reality of

implementing a welfare-to-work program

for domestic violence victims and survi-

vors in collaboration with the TANF

Department. Violence Against Women, 7(2),

211-221.

McNutt, L., Carlson, B. E., Rose, 1. M., &

Robinson, D. A. (2002). Partner violence

in the busy primary care environment.

American Journal of Preventative Medicine,

22(2), 84-91.


Nabi, R. L., & Homer, J. R. (2001). Victims

with voices: How abused women concep-

tualize the problem of spousal abuse and

implications for intervention and preven-

tion. Journal of Family Violence, 36(3),

237-253. 1

O'Keefe, M., & Mennen, F. (1998). Integrating

content on violence into a social work

pracfice curriculum. Journal of Teaching in

Social Work, 17(1/2), 81-100.

Owens-Manly, J. (1999). Battered women and

their children: A public policy response.

Afiïlia, 34(4), 439-459.

Park, Y. (2005). Culture as deficit: a critical dis-

course analysis of the concept of culture

in contemporary social work discourse.

Journal of Sociology and Social Welfare,

32(3), 13-34.

Pelto, P. J., & Pelto, G. H. (1975). Intra-cultural

diversity: Some theoretical issues.

American Ethnologist, 2, 1-18.

Personal Responsibility and Work Oppor-

tunity Recondliation Act of 1996, P. L.

104-193, 110 Stat. 2105.

Postmus, J. L. (2000). Analysis of the Family

Violence Option: A strengths perspective.

Aßlia, 15, 244-258.

Postmus, J. L. (2004). Battered and on welfare:

The experiences of women with the

Family Violence Option. Journal of

Sociology and Social Welfare, 31(2), 113-123.

Postmus, J. L., & Ortega, D. (2005). Serving

two masters: When domestic violence

and child abuse overlap. Families in

Society, 86, 483-490.

Pyles, L. (2006). Toward safety for low-income

battered women: Promoting economic

justice strategies. Families in Society, 87(1),

63-70.

Pyles, L., & Postmus, J. L. (2004). Addressing

the problem of domestic violence: How

far have we come? Aßlia. 19, 376-388.

Raphael, J. (2001). Domestic violence as a

welfare-to-work barrier. In C. M. Renzetti,

J. L. Edleson, & R. K. Bergen (Eds.),

Sourcebook on violence against women.

Thousand Oaks, CA: Sage.

Romney, A. K., Weiler, S. C, & Batchelder, W.

H. (1986). Culture as consensus: A theory

of culture and informant accuracy.

American Anthropologist, 88, 313-338.

Ross, M., & Glisson, C. (1991). Bias in sodal

work intervention with battered women.

Journal of Social Service Research, 14(3/A),

79-105.

Saunders, D. G., Holter, M. C, Pahl, L. C,

Tolman, R. M., & Kenna, C. E. (2005).

TANF workers' responses to battered

women and the impact of brief worker

training: What survivors report. Violence

Against Women, 11(21 227-254.

Shore, B. (1996). Culture in mind: Cognition, cul-

ture, and the problem of meaning. New York:

Oxford University Press.

Teddlie, C, & Tashakkori, A. (2003). Major

issues and controversies in the use of

mixed methods in the social and behav-

ioral sciences. In A. Tashakkori & C.

Teddlie (Eds.), Handbook of mixed methods

in social and behavioral research. Thousand

Oaks, CA: Sage.

Tjaden, P., & Thoennes, N. (2000). Prevalence

and consequences of male-to-female and

female-to-male violence as measured by the

National Molence Against Women Survey.

Violence Against Women, 6,142-161.

Tower, L. E. (2003). Domestic violence screen-

ing and social work education: Insti-


tutional correlates. Journal of Social Work

Education, 39, 479-494.

Weiler, S. C, & Romney, A. K. (1988). Systematic

data collection: Qualitative Research Methods

Series 10. Newbury Park, CA: Sage.

Worden, A. P., & Carlson, B. E. (2005).

Attitudes and beliefs about domestic vio-

lence: Results of a public opinion survey.

II. Beliefs about causes. Journal of Inter-

personal Violence, 20,1219-1243.

Accepted: 06/07 * ' "

Cyleste C. Collins is research assistant professor with the Center on Urban Poverty & CommunityDevelopment at Case Western Reserve University. William W. Dressier is professor at the Universityof Alabama.

Address correspondence to Cyleste C. Collins, Center on Urban Poverty & Community Development,Mandel School of Applied Social Sciences. Case Western Reserve University, 10900 Euclid Avenue,Cleveland, OH 44106-7164; e-mail: [email protected].

Associate Professor of Social WorkStephen F. Austin State UniversityNacogdoches, Texas. The associateprofessor will teach undergraduate andgradúale eourses in the B.S.W. andM.S.W. programs, advise students,participate in university, school andcommunity areas, and in scholarlyactivity. Teaching expertise required intwo of the following areas: humanbehavior aod the social enviromnent,advanced generalist practice, socialpolicy, social research, multiculturalissues, niral social work, or fieldeducation. Minimum teachingexperience required is two years,required Ph.D. in Social Woîk. This is afull-time, tenure-tmck faculty, nine-month position. Estimated sliut date:9/1/08. Salary: $53,275. Applicationinformation, submit a letter of interest,resume, transcripts and tliree letters ofrecommendation Io: Glenda Henington,Department of Human Resources,Stephen F. Austin State University. P.O.Box 13039. 201 Austin, Naeogdoches,TX 75962 (936)468-2002,ghemngton{ä)sfasu. edu

Articulo Pr on Tua Rio

Documents

Transcript of Articulo Pr on Tua Rio