How Do Children Stigmatize People With Mental Illness?

How Do Children Stigmatize People With Mental Illness?

Patrick W. Corrigan1

Joint Research Programs inPsychiatric Rehabilitation

Illinois Institute of Technology

Amy C. WatsonUniversity of Illinois at Chicago

Emeline OteyNational Institute of Mental Health

Bethesda, MD

Anne L. Westbrook,April L. Gardner, and

Theodore A. LambBiological Sciences Curriculum Study

Colorado Springs, CO

Wayne S. FentonNational Institute of Mental Health

Bethesda, MD

A way to promote eliminating stigma surrounding mental illnesses is targeting thephenomenon in children. This study’s purpose is to validate models of mental illnessstigma on children in Grades 6–8. Children completed the revised Attribution Ques-tionnaire in a pretest of a larger study on a mental health education program. Datafrom this study permitted testing of roles of demographics in these social cognitivemodels. Subsequent analyses using manifest model structural equations were mixed,but mostly showed adequate fit for multiple versions of the models. These resultssuggest that models of blame and dangerousness are relevant to the way 10 to13-year-olds stigmatize mental illness. Demographics were not found to fit thesemodels satisfactorily. Implications of these findings for stigma-change agenda arediscussed.

During the past two decades, research has begun to describe the problemswrought by mental illness stigma and to address ways to diminish it (Corrigan,2005). In partnership with advocacy groups, this research has begun to lay outa program to erase the stigma in order to increase the opportunities of peoplewith mental illness. Children often are identified as important targets forstigma change (Wahl, 2002). Perhaps the cognitive processes of elementaryschool students could be influenced so that prejudice about and discriminationtoward people with mental illness never develops or is muted. Ideally, we could

1Correspondence concerning this article should be addressed to Patrick W. Corrigan, JointResearch Programs in Psychiatric Rehabilitation, Illinois Institute of Technology, 3424 S. StateStreet, 1st Floor, Chicago, IL 60616. E-mail: [email protected]

1405

Journal of Applied Social Psychology, 2007, 37, 7, pp. 1405–1417.© 2007 Copyright the AuthorsJournal compilation © 2007 Blackwell Publishing, Inc.

mailto:[email protected]

foster future generations of adults in which the stigma of mental illness isneither so prevalent nor egregious. These kinds of programs require a betterunderstanding of how stigma develops and is maintained in children.

Theory and research on mental illness stigma has been significantlyadvanced through a translational research agenda; that is, enhancing theoreti-cal and methodological approaches to mental health issues by extrapolatingrelated ideas from basic behavioral research. Research on mental illnessstigma has borrowed heavily from basic social cognitive research that explainsthe prejudice and discrimination experienced by minority groups. The trans-lational research agenda thus far has been applied largely to understandinghow adults stigmatize people with mental illness and ways to diminish it. Thepurpose of this paper is to extrapolate and test adult models on children.

There are two models of stigmatizing attitudes that have been studied onadults: (a) Persons with serious mental illness are personally responsible fortheir symptoms/disabilities; and (b) they are dangerous and should beavoided. Weiner (1995) developed a model of causal attribution thatexplains, at least partly, the relationship between stigmatizing attitudes anddiscriminatory behavior. As outlined in Figure 1, Weiner believed that attrib-uting personal responsibility for a negative event (e.g., “That person causeshis crazy behavior”) leads to anger (“I’m sick and tired of that kind ofirresponsibility!”), diminished helping behavior (“I’m not going to give him aride”), and increased punishment (“He should be locked away in anasylum”). Conversely, attributing no blame for a harmful event (“She can’thelp herself; she’s mentally ill”) leads to pity (“That poor woman is ravagedby mental illness”) and the desire to help (“I’ll rent her a room until she’sback on her feet”). The attribution model has been validated on severalsamples (Dooley, 1995; Graham, Weiner, & Zucker, 1997; Menec & Perry,1998; Reisenzein, 1986; Rush, 1998; Steins & Weiner, 1999; Zucker &Weiner, 1993), including those specific to mental illness stigma (Corriganet al., 2002; Corrigan, Markowitz, Watson, Rowan, & Kubiak, 2003).

Responsibility attributions seem to make sense for explaining the relation-ship between mental illness stigma and discriminatory behavior. However,these kinds of attributions differ markedly from typical attitudes about mentalillness that emerge in factor analyses of public stigma; namely, that people withserious mental illness need to be segregated from society because they aredangerous (Brockington, Hall, Levings, & Murphy, 1993; Cohen & Struening,1962; Link, Monahan, Stueve, & Cullen, 1999; Pescosolido, Monahan, Link,& Stueve, & Kikuzawa, 1999). We have outlined one speculative modelelsewhere (Corrigan, 2000) and repeat it at the bottom of Figure 1. Accordingto this model, attributing a person’s behavior as dangerous leads to fear. Mostpeople respond to violent threats ofany kind with apprehension and avoidance(Johnson-Dalzine, Dalzine, & Martin-Stanly, 1996).

1406 CORRIGAN ET AL.

The purpose of the present study is to determine whether stigma modelsdeveloped for adults are explanatory for the stigma experiences of children insixth, seventh, and eighth grade. Researchers have argued that stigma, prej-udice, and discrimination are likely to be a different process in childrenbecause of their limited cognitive development. For example, research byAboud (1993, 2003) has shown that concrete operational processing is likelyto yield more discrimination in preschool children and kindergarteners thanin adults. Interestingly, as children age (e.g., 10 to 13 years old), cognitivedifferences with adults relevant to stigma seem to diminish (Corrigan,Watson, & Lahey, in press). We test this assumption by examining whetherthe path models established for adults show similar goodness of fit on middleschoolers.

These data also permitted testing of another set of hypotheses; namely, dodemographic variables describe a model with good fit and significant asso-ciations with the social cognitive models? This aspect of the models also is

Path A

Path B

Dangerous Fear

Help

Segregation

Pity

Anger

Responsibility

Avoidance

Segregation

GenderEthnicity

Grade

GenderEthnicity

Grade

Figure 1. Hypothetical paths accounting for stigmatizing reactions. The right side of Path Arepresents relationships between attributions of personal responsibility for mental illness, sub-sequent pity or anger, and the effects of this pity or anger on helping behavior or punishment(segregation). The right side of Path B represents attributions of dangerousness, subsequent fear,and avoidant behavior. The figure also describes path models where demographic variables areadded as exogenous variables.

STIGMA AND MENTAL ILLNESS 1407

illustrated in Figure 1. Previous research has examined the effects of twovariables—gender and ethnicity—on the responsibility-based model ofmental illness stigma. These studies typically viewed gender and ethnicity asexogenous variables in the model. In terms of gender, women were signifi-cantly less likely to endorse the stigma of mental illness than were men(Corrigan, Watson, & Miller, in press; Farina, 1998). In terms of ethnicity,people of color, compared to the White majority, were less likely to blameindividuals for their mental illness, more likely to sympathize with them, andless likely to avoid them in social settings (Corrigan, Backs, Green, Diwan, &Penn, 2001; Schnittker, Freese, & Powell, 2000). Hence, we expect girls fromminority samples to be less likely to endorse the stigma of mental illness.

Research also has examined whether cognitive models of responsibilityattributions and perceptions of dangerousness vary with childhood cognitivedevelopment. Although there are significant differences in development thatwould affect the social cognitive models between kindergarten and highschool (Corrigan, Watson, & Lahey, in press), we would expect to find nosuch variance in the relatively narrow years of participants in this study:children in sixth, seventh, and eighth grade.

Method

Participants were 1,566 students from 23 schools in 16 states from aroundthe country who were recruited as part of a pilot study examining the impactof a middle school curriculum entitled “The Science of Mental Illness.” Thecurriculum was developed to provide children with state-of-the-art knowledgeabout mental health and to decrease the stigma of mental illness. Teacherswere solicited (through newsletters of the National Association of BiologyTeachers and the National Science Teachers Association) to learn the curricu-lum and participate in the field study. We selected 14 teachers for the study whovolunteered their students to participate in the pilot. Only data from thepretest are reported here.

There were 1,391 children (725 female, 52.1%; 666 male) who providedsufficient responses to baseline stigma measures to be included in the analysesreported in this paper. The sample size of individual analyses was as low as1,375 because of missing data for those analyses. The subsample was drawnfrom three grades: sixth (14.5% of subsample), seventh (55.9%), and eighth(29.6%). In terms of ethnicity, 1.9% of students reported themselves as AsianAmerican, 2.4% African American, 16.2% Hispanic, 1.3% Native American,0.4% Pacific Islander, and 69.8% European American. The remaining 8.0%identified with two or more ethnic groups. Research participants weredivided into White, African American, Hispanic, and other groups for theanalysis of ethnicity effects on the social cognitive models.


Students completed the revised Attribution Questionnaire (r-AQ), aninstrument that measures the factors outlined on the right side of Figure 1. Inan earlier study on the original Attribution Questionnaire (Corrigan et al.,2003), respondents read a brief statement about Harry, a 30-year-old singleman with schizophrenia. The vignette is very brief to better represent therespondent’s reaction to the schizophrenia label, rather than other informa-tion. Respondents in the earlier study then completed 27 items on 7-pointLikert-type agreement scales. Three items were written to represent each ofthe eight factors in Figure 1 plus coercion, an additional factor not used inthis paper. The factor structure and reliability of the original AQ werevalidated in two confirmatory factor analyses (Corrigan et al., 2002, 2003).

We made two revisions to the AQ for the present study. First, the vignettewas changed to represent a child, rather than an adult: “There is a new studentin your class who just came from another school. You have heard that thisstudent has a mental illness.” Note that this is the entire vignette and, as in theoriginal AQ, was kept short purposefully to capture the participant’s reactionto the schizophrenia label. Second, the number of items was reduced to eightto generate a short test of stigma. Smaller instruments meet some of theefficiency needs imposed by survey research. Based on results from our priorconfirmatory factor analyses (Corrigan et al., 2002, 2003), the single item thatloaded most highly into each of the eight factors in Figure 1 was incorporatedinto the r-AQ. Items are provided verbatim in Table 1. All research partici-pants received r-AQ items in the order indicated in Table 1.

Results

Means and standard deviations of the eight items from the r-AQ aresummarized in the first column of Table 1 (note the small means for theresponsibility and segregation variables). The restricted range evident inthese means may limit potentially significant associations. Path analysis withmanifest variables was used to test the theoretical models outlined in the rightside of Figure 1 because it is one of the more robust measures of both the sizeand direction of associations among a set of variables. All analyses wereconducted using the SAS System’s CALIS procedure (Hatcher, 1994); theyadopted the maximum likelihood method of parameter estimation; and theywere performed on the variance–covariance matrix.

Table 1 also includes the correlation matrix for the eight variables fromwhich the variance–covariance matrix derives. Several correlation coeffi-cients were small and insignificant (e.g., between pity and anger, or betweenpity and fear): Small correlations make sense in these cases. While in the samemodel, pity and anger are represented as independent paths. Correlations like


Tab

le1

Mea

nsan

dIn

terc

orre

lati

ons

ofS

tudy

Var

iabl

es

Var

iabl

eM

SD

12

34

56

7

1.R

espo

nsib

ility

(“It

isno

tth

est

uden

t’s

faul

tif

heor

she

has

am

enta

lilln

ess.

”)

0.82

1.67

—

2.P

ity

(“I

feel

sorr

yfo

rth

ene

wst

uden

t.”)

4.91

1.83

-.16

**—

3.A

nger

(“T

hene

wst

uden

tm

akes

me

angr

y.”)

2.24

1.55

.16*

*-.

04—

4.H

elp

(“I

wou

ldhe

lpth

ene

wst

uden

t.”)

5.03

1.84

-.13

**.1

4**

-.16

**—

5.Se

greg

atio

n(“

The

new

stud

ent

shou

ldbe

lock

edin

am

enta

lhos

pita

l.”)

1.78

1.43

.16*

*-.

11**

.25*

*-.

29**

—

6.D

ange

rous

(“T

hene

wst

uden

tis

not

dang

erou

s.”)

2.26

1.64

.08*

*-.

10**

.09*

-.17

**.1

5**

—

7.F

ear

(“I

amsc

ared

ofth

ene

wst

uden

t.”)

2.32

1.64

.09*

*.0

5.2

4**

-.24

**.3

2**

.09*

*—

8.A

void

ance

(“I

will

try

tost

ayaw

ayfr

omth

ene

wst

uden

t.”)

2.84

1.74

.13*

*-.

06**

.23*

*-.

41**

.39*

*.1

1**

.33*

*

*p<

.05.

**p

<.0

1.


those represented by pity and fear are small because they come from the twoindependent models: responsibility and danger.

Goodness-of-fit indexes for the various models are presented in Table 2.The chi-square statistic that is included in this table provides a test ofthe null hypothesis that the reproduced covariance matrix has the specifiedmodel structure. Nonsignificant chi squares support good fit. The statistic,however, is very sensitive to sample size, as well as departures frommultivariate normality, and often may result in the rejection of a well-fitting model. Table 2 includes three additional fit indexes: comparativefit index (CFI), normed fit index (NFI), and non-normed fit index (NNFI;Bentler, 1989; Bentler & Bonett, 1980). Scores on these indexes varyfrom 0 to 1 and are considered to support fit when they are greater than.90. Finally, Table 2 includes standardized coefficients representing the sizeof the association of individual paths within each model. Significant tvalues (t > 1.96, p < .05) for individual coefficients are marked with anasterisk.

Responsibility

Table 2 summarizes two responsibility path models, predicting help andsegregation, respectively. Chi-square statistics for both models did notsupport a good fit for the models. However, the other fit indicators largelysupported the models. CFI and NFI were greater than .90 for both models,while NNFI approached .90 for the help model. As outlined on the right sideof Table 2, standardized coefficients for the elements of all models weresignificant, and corresponding standard errors were satisfactory. Willingnessto help was positively associated with pity and negatively with anger(R2 = .09). Responsibility inversely predicted pity and directly predictedanger. Endorsing segregation was positively associated with anger and nega-tively with pity (R2 = .13).

Dangerousness

Table 2 also provides two versions of the dangerousness model predictingavoidance and segregation. Similar to the findings already reported, fit indi-cators were mixed for avoidance. Chi square did not support a good fit, butCFI and NFI did, and NNFI approached .90. The two standardized pathcoefficients were significant, and standard error was appropriate. Fear waspositively associated with avoidance (R2 = .11). Danger predicted fear,though at a fairly low level (R2 = .01). Note that none of the fit indicatorssupported the model for danger, fear, and segregation.


Tab

le2

Sum

mar

yof

SE

Mfo

rV

ario

usB

SC

SM

odel

s

Mod

el

Fit

indi

cato

rsP

ath

coef

ficie

nt

pfo

rc2

c2 /df

CF

IN

NF

IN

FI

Pat

hSt

anda

rdiz

edco

effic

ient

Pat

hSt

anda

rdiz

edco

effic

ient

Pat

hSt

anda

rdiz

edco

effic

ient

Pat

hSt

anda

rdiz

edco

effic

ient

R→

P→

H↓ A

n

.001

6.58

.958

.872

.951

R–

P-.

19*

R–A

n.2

3*P

–H

.11*

An

–H

-.27

*

R→

P→

S↓ A

n

.001

14.3

8.9

20.7

59.9

15R

–P

-.19

*R

–An

.22*

P–

S-.

08*

An

–S

.35*

D→

F→

Av

.003

8.77

.956

.869

.951

D–

F.0

9*F

–Av

.33*

D→

F→

S.0

001

22.5

8.8

77.6

32.8

73D

–F

.10*

F–S

.31*

Not

e.SE

M=

stru

ctur

aleq

uati

onm

odel

ing;

BSC

S=

Bio

logi

cal

Scie

nces

Cur

ricu

lum

Stud

y;C

FI

=co

mpa

rati

vefit

inde

x;N

NF

I=

non-

norm

edfit

inde

x;N

FI

=no

rmed

fitin

dex;

R=

resp

onsi

bilit

y;P

=pi

ty;

H=

help

;S

=se

greg

atio

n;A

v=

avoi

d;D

=da

nger

;A

n=

ange

r;F

=fe

ar.

The

dow

nwar

d-po

inti

ngar

row

sre

pres

ent

the

mod

elas

pres

ente

din

Pat

hA

ofF

igur

e1:

resp

onsi

bilit

yle

adin

gto

feel

ings

ofei

ther

pity

oran

ger

and

cons

eque

ntly

toei

ther

help

ing

orse

greg

atio

nbe

havi

ors.

*tva

lue

for

coef

ficie

ntex

ceed

s1.

96,p

<.0

5.


Effects of Gender, Ethnicity, and Grade

A second important set of questions for this study focused on demo-graphics as moderator variables of the social cognitive models. We added threedemographic variables—gender, ethnicity, and grade level—as exogenousvariables in Paths A and B (left side of Figure 1). The structural equations forresponsibility failed to reach good fit when all three demographic variableswere included in the structural equations. Goodness-of-fit indicators rangedbetween .59 and .78, and chi square was highly significant ( p < .001).

We expected that the narrow grade level of participants in this studywould not provide a sufficiently broad range of cognitive development, sograde level would not fit well in the model. The results support this expecta-tion: The standardized estimate representing the relationship between gradelevel and responsibility (.04) was not significant ( p > .10). We reran the fitindicators with grade level removed from the equation. Once again, themodel combining social cognitive constructs and demographics (in this case,gender and ethnicity) was not significant. Fit indicators ranged from .62 to.80, and chi square was significant ( p < .001). None of the other models ondangerousness with demographic variables showed good fit. Hence, weconclude that gender and ethnicity failed to fit well in our models.

Discussion

The present study examined whether two models that describe the stigmaof mental illness in adults also apply to children. Results were mixed, butmostly supported the comparison. Children who viewed other children asresponsible for their mental illness expressed more anger and less pity towardthem, which, in turn, was related to being less willing to help them and toendorsing treatment in segregated settings. This model is consistent with theassertions of Weiner’s (1995) attribution theory. In addition, children whoviewed people with mental illness as dangerous were likely to be fearful ofthem and to try to avoid them. Note that all of the standardized estimatesrepresenting specific relationships in the two models were significant. Hence,this study shows some support for adult models of the stigmatizing processesas applied to sixth, seventh, and eighth graders.

Findings like these have implications for developing programs meant todiminish stigma. For example, avoidance is an especially important variablein understanding behavior in children. It may explain why some people whoare labeled as different because of mental illness are ostracized by their peers.Hence, antistigma programs wishing to enhance the quantity and quality ofinteractions between children who are labeled mentally ill and their peerswould benefit from these kinds of findings. Most of the research in this area


targets the prejudice related to ethnicity. Therefore, our conclusions here arebased on extrapolation of these findings.

Aboud and colleagues (Aboud & Doyle, 1996; Aboud & Fenwick, 1999)hypothesized that peers might be better resources for stigma change becausethey are less likely to censor opinions about a topic and because the cognitivestyle of communication is automatically at the level of a same-age peer. Atypical format for this kind of strategy is to pair a low-prejudiced child witha same-age, familiar, high-prejudiced child. The dyad is then presented witha stimulus picture (e.g., for racial prejudice, separate pictures of a Black childand a White child), asked to rate the person in the picture across a series ofpositive (the child is neat, honest, a nice person) and negative (the child is abully, lazy, and dirty) values. Of more importance, the dyad is asked todiscuss their rationale for each rating.

Interestingly, observers of these dyads have noted that neither the high-nor the low-prejudiced child attempted to dissuade his or her partner in termsof rating or rationale. Of further note, although it was clear at the beginningof these discussions who was the high- and low-prejudiced child, the dyadappeared quite similar in response by the end of the discussion (Aboud &Doyle, 1996; Aboud & Fenwick, 1999). More balanced evaluation ofin-groups and out-groups corresponded with descriptions of Whites withmore negative attributes and Blacks with more positive features.

Other research on peer collaboration has shown that the solutions gener-ated by participants are more mature than their initial positions (Chapman &McBride, 1992). Specific qualities of the interaction seem to facilitate positivegains. For example, expression of contradictory positions along with anexplanation is better than agreement and is better than no explanation(Nelson & Aboud, 1985). Interpretive statements that exceed mere descrip-tion are instrumental in helping dyads to achieve more balanced statements(Teasley, 1995; Ticao & Aboud, 1998). Future research must adapt thesefindings to change children’s attitudes about mental illness.

The present study examined the inclusion of demographics in the socialcognitive model. Results failed to find good fit in the models that incorpo-rated grade, ethnicity, and gender. We argued that students in Grades 6through 8 are fairly homogeneous in terms of cognitive development. Thislimited variance would fail to yield significant grade effects in the model. Inour previous research, ethnicity represented mostly African Americans (Cor-rigan et al., 2001; Schnittker et al., 2000). Non-White research participants inthe present study were largely Hispanic (53.6% of the non-White sample).Cultural differences between Hispanics and African Americans may explainthe absence of significant effects for ethnicity. Moreover, language differ-ences between African Americans and Hispanics may have resulted in thelack of support for ethnicity in our models.


The lack of good fit for the model with gender is more difficult to explain.In a comprehensive review of studies on gender and mental illness stigma,Farina (1998) found no significant effects between men and women in somestudies. The absence of a significant model fit in our study replicates some ofFarina’s conclusions.

The present study was limited by the relatively small R2 describing thesize of the relationship between endogenous variables and other elements ofthe path. Hence, the models used in this study were limited in their expla-nation of discriminatory behaviors and accounted for significantly less vari-ance than those found in the adult samples (Corrigan et al., 2002, 2003). Asalluded to previously, perhaps this difference is a result of the participantsbeing children. Absent from the present study was assessment of cognitivestage and how it may have influenced endorsing aspects of the stigmamodel. In particular, we hypothesize that development of abstract cognitiveprocesses will be an especially important mediator of stigma. This is aninteresting construct that must be integrated into future studies.

References

Aboud, F. E. (1993). The developmental psychology of racial prejudice.Transcultural Psychiatric Research Review, 30, 229–242.

Aboud, F. E. (2003). The formation of in-group favoritism and out-groupprejudice in young children: Are they distinct attitudes? DevelopmentalPsychology, 39, 48–60.

Aboud, F. E., & Doyle, A. B. (1996). Parental and peer influences on chil-dren’s racial attitudes. International Journal of Intercultural Relations, 20,371–383.

Aboud, F. E., & Fenwick, V. (1999). Exploring and evaluating school-basedinterventions to reduce prejudice. Journal of Social Issues, 55, 767–785.

Bentler, P. M. (1989). EQS structural equations program manual. LosAngeles: BMDP Statistical Software.

Bentler, P. M., & Bonett, D. G. (1980). Significance tests and goodness of fitin the analysis of covariance structures. Psychological Bulletin, 88, 588–606.

Brockington, I. F., Hall, P., Levings, J., & Murphy, C. (1993). The com-munity’s tolerance of the mentally ill. British Journal of Psychiatry, 162,93–99.

Chapman, M., & McBride, M. L. (1992). The education of reason: Cognitiveconflict and its role in intellectual development. In C. U. Shantz & W. W.Hartup (Eds.), Conflict in child and adolescent development (pp. 36–39).New York: Cambridge University Press.


Cohen, J., & Struening, E. L. (1962). Opinions about mental illness in thepersonnel of two large mental hospitals. Journal of Abnormal Psychology,64, 349–360.

Corrigan, P. W. (2000). Mental health stigma as social attribution: Implica-tions for research methods and attitude change. Clinical Psychology:Science and Practice, 7, 48–67.

Corrigan, P. W. (2005). Changing stigma through contact. Advances inSchizophrenia and Clinical Psychiatry, 1, 54–58.

Corrigan, P. W., Backs, A. E., Green, A., Diwan, S. L., & Penn, D. L. (2001).Prejudice, social distance, and familiarity with mental illness. Schizo-phrenia Bulletin, 27, 219–226.

Corrigan, P. W., Markowitz, F. E., Watson, A. C., Rowan, D., & Kubiak,M. A. (2003). Attribution model of public discrimination towardspeople with mental illness. Journal of Health and Social Behavior, 44,162–179.

Corrigan, P. W., Rowan, D., Green, A., Lundin, R., River, P., Uphoff-Wasowski, K., White, K., & Kubiak, M. A. (2002). Challenging twomental illness stigmas: Personal responsibility and dangerousness. Schizo-phrenia Bulletin, 28, 293–310.

Corrigan, P. W., Watson, A. C., & Lahey, B. B. (in press). How childrenstigmatize people with mental illness. Clinical Psychology Review.

Corrigan, P. W., Watson, A. C., & Miller, F. E. (in press). The impact ofmental illness and substance abuse stigma on family members. Journal ofFamily Psychology.

Dooley, P. A. (1995). Perceptions of the onset controllability of AIDS andhelping judgments: An attributional analysis. Journal of Applied SocialPsychology, 25, 858–869.

Farina, A. (1998). Stigma. In K. T. Mueser & N. Tarrier (Eds.), Handbookof social functioning in schizophrenia (pp. 247–279). Boston: Allyn &Bacon.

Graham, S., Weiner, B., & Zucker, G. S. (1997). An attributional analysis ofpunishment goals and public reactions to O. J. Simpson. Personality andSocial Psychology Bulletin, 23, 331–346.

Hatcher, L. (1994). A step-by-step approach to using the SAS system for factoranalysis and structural equation modeling. Cary, NC: SAS Institute, Inc.

Johnson-Dalzine, P., Dalzine, L., & Martin-Stanly, C. (1996). Fear of cri-minal violence and the African American elderly: An assessment of acrime prevention strategy. Journal of Negro Education, 65, 462–469.

Link, B. G., Monahan, J., Stueve, A., & Cullen, F. T. (1999). Real in theirconsequences: A sociological approach to understanding the associationbetween psychotic symptoms and violence. American Sociological Review,64, 316–332.


Menec, V. H., & Perry, R. P. (1998). Reactions to stigmas among Canadianstudents: Testing an attribution–affect–help judgment model. Journal ofSocial Psychology, 138, 443–453.

Nelson, J., & Aboud, F. E. (1985). The resolution of social conflict betweenfriends. Child Development, 56, 1009–1017.

Pescosolido, B. A., Monahan, J., Link, B. G., Stueve, A., & Kikuzawa, S.(1999). The public’s view of the competence, dangerousness, and need forlegal coercion of persons with mental health problems. American Journalof Public Health, 89, 1339–1345.

Reisenzein, R. (1986). A structural equation analysis of Weiner’s attribution–affect model of helping behavior. Journal of Personality and SocialPsychology, 50, 1123–1133.

Rush, L. (1998). Affective reactions to multiple social stigmas. Journal ofSocial Psychology, 138, 421–430.

Schnittker, J., Freese, J., & Powell, B. (2000). Nature, nurture, neither, nor:Black–White differences in beliefs about the causes and appropriate treat-ment of mental illness. Social Forces, 78, 1101–1132.

Steins, G., & Weiner, B. (1999). The influence of perceived responsibility andpersonality characteristics on the emotional and behavioral reactions topeople with AIDS. Journal of Social Psychology, 139, 487–495.

Teasley, S. D. (1995). The role of talk in children’s peer collaboration.Developmental Psychology, 3, 207–220.

Ticao, C. J., & Aboud, F. E. (1998). A problem-solving approach to nutritioneducation with Filipino mothers. Social Science and Medicine, 46, 1531–1541.

Wahl, O. F. (2002). Children’s views of mental illness: A review of theliterature. Psychiatric Rehabilitation Skills, 6, 134–158.

Weiner, B. (1995). Judgments of responsibility: A foundation for a theory ofsocial conduct. New York: Guilford.

Zucker, G. S., & Weiner, B. (1993). Conservatism and perceptions ofpoverty: An attributional analysis. Journal of Applied Social Psychology,23, 925–943.


Documents

How Do Children Stigmatize People With Mental Illness?