race/ethnicity and inter-informant agreement in …...highly valued, reports from adolescents,...

J O U R N A L O F E M O T I O N A L A N D B E H A V I O R A L D I S O R D E R S , F A L L 2 0 0 4 , V O L . 1 2 , N O . 3 , P A G E S 1 4 5 – 1 5 6 145

Race/Ethnicity and Inter-InformantAgreement in Assessing Adolescent

Psychopathology

ANNA S. LAU, ANN F. GARLAND, MAY YEH, KRISTEN M. MCCABE,PATRICIA A. WOOD, AND RICHARD L. HOUGH

Although the use of multiple informantsin assessing youth psychopathology ishighly valued, reports from adolescents,parents, and teachers consistently yieldlittle cross-informant agreement (Achen-bach,McConaughy,& Howell,1987). Thismodest rate of inter-informant agreementmay be expected, given the situationalspecificity associated with key informantreports. Although youth self-reports mayreflect behaviors that occur across a va-riety of settings and situations (home,school, with peers), as well as internal ex-periences that cannot be observed, reportsby adult informants are typically based onobservations in more circumscribed con-ditions (either home or school). Giventhese different samplings of behaviors, in-stances of disagreement are not surpris-ing. Research suggests that levels of con-cordance between informants varies byclinical variables, such as the type ofsymptoms assessed (Achenbach et al.,1987), the severity of family dysfunction(Kolko & Kazdin, 1993), and the presenceof parental psychopathology (Frick, Sil-verthorn, & Evans, 1994). Identifying con-ditions under which agreement is particu-

larly poor can help highlight circum-stances in which a multimethod, multi-informant approach to clinical assessmentis essential. Moreover, studying the pre-dictors of inter-informant (dis)agreementcan illuminate factors that influence theidentification of youth psychopathology.

Racial/ethnic variation in inter-informant agreement in behavior problemratings is a central issue in cross-culturalassessment research that seeks to under-stand whether scores on a given ratingscale mean the same thing across differ-ent cultural groups. Cross-cultural equiv-alence in psychological assessment is

analogous to interrater reliability (Reid,1995). For both interrater reliability andcross-cultural equivalence to be demon-strated, four conditions must be met(Marsella & Kameoka, 1989). First, lin-guistic equivalence and conceptual equiv-alence require that raters share a commonunderstanding of the attribute being ratedas well as an understanding of behaviorsthat are representative of that attribute.Next, metric equivalence requires thatraters share a common metric in order toaccurately scale behaviors relevant to theattribute being rated. Finally, normativeequivalence demands that raters are cap-

The authors examined the influence of race/ethnicity on patterns of ratings of adolescent psy-

chopathology completed by adolescents, parents, and teachers in a sample of 600 adolescents.

Robust racial/ethnic differences in behavior problems emerged with parent and teacher reports,

but not with adolescent self-reports. Discrepancy scores revealed that minority parents reported fewer

behavior problems than did youth.Such findings have important implications for treatment seeking, as par-

ents are often the initiators of service use for youth. Compared to other informants, teachers reported

fewer internalizing problems among minority adolescents in general, fewer externalizing problems in

Asian/Pacific Islanders, and more externalizing problems for African Americans.These findings are consis-

tent with a number of explanations, including cultural differences in adult distress thresholds, racial/

ethnic bias among teachers, and cultural differences in symptom expression.

146 J O U R N A L O F E M O T I O N A L A N D B E H A V I O R A L D I S O R D E R S , F A L L 2 0 0 4 , V O L . 1 2 , N O . 3

able of determining the occurrence andnonoccurrence of behaviors relevant tothe attribute being rated. Accordingly, ev-idence of racial/ethnic differences in inter-informant agreement on ratings of youthbehavior problems is suggestive of prob-lems with cross-cultural linguistic, con-ceptual, metric, or normative equivalence.

Indeed, by examining the influence ofrace/ethnicity on inter-informant agree-ment and discrepancies in ratings of youthpsychopathology, we can generate hy-potheses about cultural differences in de-fining and reporting symptoms and racial/ethnic biases in assessment. Racial/ethnicdifferences in informant agreement mayreflect a lack of cross-cultural conceptualequivalence that is driven by differentialattention to, interpretation of, and toler-ance for adolescent behaviors. As Weiszet al. (1997) pointed out, the study ofyouth psychopathology “is inevitably thestudy of two phenomena: the behavior ofthe child, and the lens through whichadults view child behavior” (p. 569). Al-though culture can influence the actualexpression of child psychopathology, theadult distress threshold hypothesis con-tends that culture also influences adults’attitudes toward child behavior and the de-termination of whether a behavior consti-tutes a distressing problem. Findings ofinformant disagreement associated withrace/ethnicity may indicate cultural vari-ation in distress thresholds for judgingyouth behaviors as problematic.

It is also important to remember thatthe study of race/ethnicity encompassesnot only the study of cultural variation butalso processes related to minority sta-tus and associated sociocontextual fac-tors. Systematic variation in informant(dis)agreement by youth race/ethnicitymay also be explained by the influencesof racial/ethnic stereotypes and biases.The perceptions of extrafamilial adult ob-servers, such as teachers, may be particu-larly subject to the influence of race-related beliefs about base rates of youthdeviant and prosocial behaviors. Researchindicates that teacher expectancies for stu-dent behavior and achievement have his-torically been related to student race (Du-sek & Joseph, 1983). Studies have shownthat teachers judge African American chil-

dren as having more disruptive or hyper-active symptoms, poorer future educa-tional prognoses (Pigott & Cowen, 2000),and lower social competence (Lethermon,Williamson, Moody, & Wozniak, 1986)than Caucasian children. Teacher re-sponse to misbehavior tends to be moresevere for transgressions of AfricanAmerican students (Marwit, 1982), andteachers refer African American youth fortreatment of behavior problems more fre-quently than do African American parents(Gottlieb, Gottlieb, & Trongone, 1991).Race appears to be an influential factor indecisions to place children in special ed-ucation for emotional problems (Prieto &Zucker, 1981), where African Americanstudents are overrepresented (McCabe et al., 1999; Oswald, Coutinho, Best, &Singh, 1999). In contrast, teachers havebeen found to view Asian American youthas model students who are respectful anddiligent but passive, lacking assertivenessand social competence (Bannai & Cohen,1985; Schneider & Lee, 1990). Consistentwith this stereotype, Chang and Sue(2003) found that teachers rate overcon-trolled behaviors (e.g., shyness, worry) asmore typical for Asians than for Cau-casians and African Americans. This lit-erature suggests that patterns of dis-crepant teacher ratings may be influencedby race-related beliefs and behavioral ex-pectations.

Three studies of adolescent boys haveexamined whether inter-informant dis-crepancies in ratings of behavior prob-lems emerge differently depending on therace/ethnicity of the youth being assessed.Using data from the Pittsburgh YouthStudy (PYS), Fabrega, Ulrich, and Loeber(1996) studied cross-informant differencescores in a sample of 290 African Ameri-can and 216 Caucasian boys. Adolescent,parent, and teacher ratings on the variousforms of the Achenbach checklists (ChildBehavior Checklist [CBCL; Achenbach,1991b], Teacher Report Form [TRF;Achenbach, 1991c], Youth Self-Report[YSR; Achenbach, 1991d]) were obtainedto examine the influence of race/ethnicityon inter-informant differences in ratingsof psychopathology. Results revealed nomain effect of race/ethnicity on adolescent–parent difference scores. However, on at-

tention problems, aggression, and delin-quency, teachers rated African Americanstudents as having more problems thanCaucasians compared to youth and parentreports.

Also using data from the PYS,Young-strom, Loeber, and Stouthamer-Loeber(2000) extended these findings while con-trolling for multiple correlates of inter-informant agreement. These authors op-erationalized agreement using multipledefinitions and found that the effect ofrace on agreement persisted even whensocioeconomic status (SES), and care-giver depression, stress, substance abuse,and antisocial behavior were taken intoaccount. Consistent with the previous re-port, race was associated with increaseddisagreement between teachers and par-ents and teachers and youth for exter-nalizing, but not internalizing, problems.Teachers reported more externalizing prob-lems in African American students thandid parents and adolescents.

Zimmerman, Khoury, Vega, Gil, andWarheit (1995) examined teacher and par-ent perceptions of behavior problems in asample of 236 African American, His-panic, and Caucasian middle school boysin Dade County, Florida. Consistent withfindings from the PYS, African Americanstudents received higher problem scoresfrom their teachers than from their parentsand were more likely than Hispanic andCaucasian students to be classified as clin-ical cases by their teachers than by theirparents.

The literature reviewed provides pre-liminary evidence that inter-informantdiscrepancies and agreement vary sys-tematically by race/ethnicity. Notably,adolescent–teacher and parent–teacherdiscrepancies may be attributable in partto cultural differences or racial/ethnicbiases. Teacher disagreements with ado-lescents and/or parents are more pro-nounced for African Americans than forCaucasians (Fabrega et al., 1996; Young-strom et al., 2000) and Hispanics (Zim-merman et al., 1995). Although the pre-vious studies found no racial/ethnicdifferences in adolescent–parent discrep-ancies, it is possible that including racial/ethnic groups with a wider range of in-tergenerational differences in cultural

J O U R N A L O F E M O T I O N A L A N D B E H A V I O R A L D I S O R D E R S , F A L L 2 0 0 4 , V O L . 1 2 , N O . 3 147

orientation or acculturation (i.e., groupsthat include high proportions of immi-grant families) will yield different results.Youth in immigrant families tend to adoptthe values and behaviors of the host cul-ture more readily than do their parents(Szapocznik & Truss, 1978); parent–youth disagreements about youth problembehaviors may be more pronounced infamilies with intergenerational culturalgaps.

The current study will extend the liter-ature in three ways. First, the inclusion oftwo minority groups (Hispanic and AsianPacific/Islander) primarily composed ofimmigrants in addition to Caucasians andAfrican Americans may provide a morethorough understanding of cultural influ-ences on intergenerational (dis)agreementon youth behavior problems. Second, weexamined inter-informant agreement re-garding youth behavior problems in ahigh-risk sample of youth presenting inpublic sectors of care. The study of infor-mant discrepancies and agreement in thispopulation may be relevant to mentalhealth assessment, referral, and servicereceipt in community settings than stud-ies of general community samples. Third,this study was the first to examine ethnic-ity and inter-informant discrepancies andagreement in a sample that included boysand girls. The two main objectives of thestudy were (a) to compare levels of be-havior problems according to self-reports,parent reports, and teacher reports amongCaucasian, African American, Hispanic,and Asian/Pacific Islander adolescentssampled from public services and (b) toexamine the effects of race/ethnicity onadolescent–parent,adolescent–teacher,andparent–teacher discrepancies and agree-ment in ratings of adolescent problems. Inaddition, we were able to explore potentialinteractions between race/ethnicity andgender on inter-informant discrepanciesand agreement.

Based on the previous literature re-viewed, we expected race/ethnicity to beassociated with distinct patterns of infor-mant discrepancies in reports of youthbehavior problems. Specifically, we de-veloped the following hypotheses: (a) com-pared to other racial/ethnic groups, Afri-can American youth would be rated as

having more externalizing problems byteachers than by youth or parents and (b) compared to other groups, Asian/Pacific Islander youth would be rated ashaving fewer externalizing and internaliz-ing problems by teachers than by youth orparents. In addition, we predicted that anal-yses of inter-informant agreement wouldyield (a) lower rates of youth–parentagreement about behavior problems infamilies at heightened risk of intergener-ational acculturation gaps, namely Asian/Pacific Islander and Hispanic families,and (b) lower rates of parent–teacher agree-ment for African American and Asian/Pacific Islander adolescents than for Cau-casian adolescents.

METHOD

Participants

The Patterns of Youth Mental Health Carein Public Service Systems is an epidemi-ologic study of youth served in five pub-lic systems of care in San Diego County:mental health (MH), juvenile justice (JJ),child welfare (CW), alcohol and drug ser-vices (AD), and special education ser-vices for youth with serious emotionaldisturbance (SED). The sampling framewas composed of a complete enumerationof the 12,662 youth 6 to 17 years old whowere served in at least one of these fivesectors in the last half of Fiscal Year (FY)1996–1997. From this population, a rep-resentative stratified random sample of1,715 youth was surveyed. (For more de-tailed information on sampling and re-cruitment, see Garland et al., 2001.) Strat-ification variables were race/ethnicity,level of restrictiveness of care, and sectoraffiliation. A poststratification weightingprocedure (Henry, 1990) was used to en-sure that the data would be representativeof the population of youth in service sys-tems in San Diego.

Youth ages 11 to 17 years (M = 14.9,SD = 2.2) who identified themselves asCaucasian, African American, Hispanic,or Asian/Pacific Islander, and for whomself-, parent, and teacher reports of psy-chopathology were available, were se-lected for inclusion in this study. The ageand race/ethnicity criteria yielded a sam-

ple of 1,131 eligible adolescents (65.9%of survey sample). Exclusion of adoles-cents with one or more missing informantreport on the behavior problem scalesyielded a final sample of 600 adolescents(53% of eligible adolescents), including254 Caucasians (55.9% of eligible Cau-casians), 136 African Americans (54.4%of eligible African Americans), 163 His-panics (49.2% of eligible Hispanics), and47 Asian/Pacific Islanders (49% of eli-gible Asian/Pacific Islanders), with 67% of the sample being male. Race/ethnicity,age, gender, and involvement in the MH,AD, SED, and CW sectors were unrelatedto exclusion from the sample due to miss-ing measures. However, involvement inthe JJ sector was related to greater likeli-hood of exclusion due to missing mea-sures (β = −.46, p = .005). Immigrantfamilies composed the majority of theHispanic and Asian/Pacific Islandergroups (59% and 61% foreign-born par-ents, respectively). In terms of sector af-filiation (in FY 1996–1997), 68 youthwere active in AD, 334 were active in MH,181 were active in SED, 165 were activein JJ, and 135 were active in CW (37.6%of participants had an active case in morethan one sector). At the time of the inter-view, 17.7% of the sample was residing in out-of-home placements (5.1% fostercare, 2.3% group homes, 4.3% psychiatrichospital or residential treatment center,6.0% incarcerated). Parent ratings wereprovided by a biological parent in 76.2%of cases, by a step/adoptive/foster parentin 22% of cases, and by a professionalcaregiver/service provider in the remain-ing 1.8% of cases.

Sector affiliation was associated withrace/ethnicity such that African Ameri-can, Asian/Pacific Islander, and Hispanicadolescents were more likely than Cau-casian adolescents to be active in JJ, χ2(3,N = 600) = 20.05, p < .001, and AfricanAmerican adolescents were more likely tobe active in CW than Caucasian, Hispanic,and Asian/Pacific Islander youth, χ2(3,N = 600) = 12.64, p = .005. Because youthinvolved with services related to alcohol,drugs, and/or mental health (ADM; MH,AD, and SED sectors) have higher ratesof behavior problems across informant re-ports than do youth in other systems of


care (CW and JJ sectors), we included acontrol variable denoting affiliation in oneof the ADM sectors in our analyses ofracial/ethnic effects on assessments ofadolescent psychopathology.

Family income was measured with anincremental scale that allowed partici-pants to select a value (range = 1–32) thatcorresponded with distinct annual levelsof income from $1,000 or less to $200,000or more. The median income range for thissample was $20,000 to $24,999. Family in-come was associated with race/ethnicity(F = 9.84, p < .001) such that Caucasianfamilies had a higher mean incomerange (M = $19,000–$19,999, SD = 6.89)than African American (M = $16,000–$16,999, SD = 6.82), Hispanic (M =$14,000–$14,999, SD = 6.48), and Asian/Pacific Islander families (M = $16,000–$16,999, SD = 7.65).

Measures

Adolescent, parent, and teacher ratings ofpsychopathology on the Achenbach be-havior rating scales (CBCL, YSR, andTRF) were obtained. In the current sam-ple, the broadband internalizing and ex-ternalizing scales from the YSR, CBCL,and TRF demonstrated good internal con-sistency across all racial/ethnic groups.Cronbach alphas ranged from .86 to .94for Caucasians, from .87 to .95 for AfricanAmericans, from .88 to .96 for Hispanics,and from .89 to .95 for Asian/Pacific Is-landers. In terms of validity, the internal-izing and externalizing scales on the YSR,CBCL, and TRF were significantly corre-lated for all racial/ethnic groups with par-ent reports of youth impairment on the Co-lumbia Impairment Scale (CIS; Bird et al.,1993), a 13-item continuous scale that re-liably assesses the extent to which a childhas experienced functional impairmentover the previous 6 months.

Raw scores on the cross-informant in-ternalizing and externalizing broadbandscales were examined (Achenbach, 1991a),with higher scores indicating more be-havior problems. The CBCL, YSR, andTRF contain 89 (of 112) identical items.We used a common-items approach to ex-amine cross-informant agreement, to en-sure that differing item content was not

contributing to disagreements. As a result,the cross-informant broadband scores ex-cluded up to seven informant-specificitems that would otherwise contribute to the broadband scores individually de-rived from the CBCL, TRF, and YSR.Inter-informant agreement was examinedacross three dyads (adolescent–parent,adolescent–teacher, parent–teacher). Fol-lowing Youngstrom et al.’s (2000) ap-proach, we evaluated distinct indices ofagreement, differences in levels of rawscores, and interrater qcorrelations. Cross-informant difference scores (e.g., discrep-ancies between adolescents [YSR] andparents [CBCL]) were computed by sub-tracting the raw score provided by the sec-ond informant (e.g., parent [CBCL]) fromthe raw score of the first informant (e.g.,adolescent [YSR]). Interrater q correla-tions were simply the Pearson correlationsbetween the sets of items provided by the two informants. These two indices ofagreement complement one another be-cause difference scores provided a metricindicating which informant reported moreproblems: Positive numbers indicated thatthe first informant in the dyad reportedmore problems than the second informant.Difference scores are sensitive to the lev-els but not to the shape or dispersion ofprofile scores (Youngstrom et al., 2000).In contrast, the q correlations are not sen-sitive to differences in the levels of prob-lems reported by the different informants,but they convey information about theshape and dispersion of the profile of itemscores (Waller & Meehl, 1998). For ex-ample, two informants may agree aboutthe overall level of problems withoutdemonstrating a strong pattern of agree-ment on specific constituent items. Dif-ference scores capture the extent to whichthe dyad agreed on the overall level of symptoms, and q correlations reflectagreement on symptom pattern. There-fore, by examining both types of indiceswe can determine which independent va-riables influence each aspect of agree-ment.

Analyses

MANCOVAs were conducted to examinethe effects of race/ethnicity on the cross-

informant broadband scale scores fromeach measure, on difference scores be-tween the pairs of informants, and on q correlations between pairs of infor-mants. Gender and the interaction be-tween gender and race/ethnicity were alsoexamined as independent variables. Co-variates included youth age, gender, fam-ily income, and affiliation with ADM sec-tors of care. It was important to accountfor youth demographics in our analysesbecause raw scores on these behaviorproblem scales do not take into accountage and gender norms. Furthermore, itwas important to control for family in-come in our analyses because it is oftenargued that racial/ethnic differences inyouth adjustment may be accounted for bysocioeconomic disadvantage. Finally, aspreviously discussed, we controlled forsector affiliation because race/ethnicitywas associated with ADM involvementand ADM involvement was associatedwith increased behavioral problems. Anal-yses were adjusted for sampling weightsusing SPSS version 10.0 (see Note).

RESULTS

Racial/Ethnic Differences onBehavior Problems, by Informant

Figure 1 displays the mean level of be-havior problems reported by each infor-mant, broken down by race/ethnicity. Thepreliminary multivariate test revealed sig-nificant effects of race/ethnicity, F(18,1665) = 3.25, p < .001; gender, F(6,553) = 11.72, p < .001; age, F(6, 553) =7.27, p < .001; family income, F(6,553) = 2.90, p < .001; and involvement inan ADM sector, F(6, 553) = 6.84, p < .001.There was no significant multivariateeffect of the interaction between genderand race/ethnicity on the cross-informantbroadband scales. In the following sec-tions, follow-up univariate ANCOVA re-sults are presented, focusing on the effectsof race/ethnicity on behavior problems foreach informant broadband scale.

Adolescent Reports. After covaryingfor gender, age, family income, and ADM


sector affiliation, race/ethnicity was notsignificantly associated with youth-reported internalizing problems or exter-nalizing problems.

Parent Reports. After covarying forgender, age, family income, and ADMsector affiliation, results indicated thatCBCL internalizing scores were associ-ated with race/ethnicity, F(3, 572) = 4.26,p = .005, such that Caucasian adolescentsreceived significantly higher scores thanAfrican American adolescents (TukeyHSD = 3.32, p = .005). CBCL externaliz-ing scores were significantly associatedwith race/ethnicity, F(1, 573) = 3.06, p =.028, such that Caucasians received higherscores than African Americans (TukeyHSD = 2.82, p = .048), Hispanics (TukeyHSD = 2.74, p = .034), and Asian/PacificIslanders (Tukey HSD = 4.64, p = .035).

Teacher Reports. With covariates heldconstant, race/ethnicity was associatedwith TRF internalizing scores, F(3, 569) =3.45, p = .016, such that Caucasians re-ceived higher scores than African Ameri-cans (Tukey HSD = 2.34, p = .029). TRFexternalizing scores were also signifi-

cantly related to race/ethnicity, F(3,569) = 4.84, p = .002, with African Amer-icans receiving higher scores than Cau-casians (Tukey HSD = 2.63, p = .036) andAsian/Pacific Islanders receiving lowerscores than African Americans (TukeyHSD = 7.21, p = .001) and Caucasians(Tukey HSD = 4.89, p = .039).

Racial/Ethnic Differences inCross-Informant Discrepancies

Figure 2 displays the estimated mean dif-ference scores (adjusted for covariates)between the three dyads of interest, byrace/ethnicity for each problem type. Thepreliminary multivariate test indicatedthat race/ethnicity, F(12, 1671) = 4.33,p < .001; gender, F(4, 555) = 9.15, p <.001; age, F(4, 555) = 9.80, p < .001; fam-ily income, F(4, 555) = 4.31, p = .002; andADM sector involvement, F(4, 555) =3.42, p = .009, were significantly associ-ated with the multivariate set of inter-informant difference scores, but the inter-action between race/ethnicity and genderwas not. The following sections presentfollow-up univariate ANCOVA results for

the independent variable of race/ethnicityon each dyad broadband difference score.

Adolescent–Parent Difference Scores.After controlling for gender, age, ADMsector affiliation, and family income, in-ternalizing YSR–CBCL difference scoreswere significantly associated with race/ethnicity, F(3, 568) = 5.42, p = .001,such that Caucasians had lower differencescores than African Americans (TukeyHSD = 4.82, p < .001), Hispanics (TukeyHSD = 2.82, p = .033), and Asian/PacificIslanders (Tukey HSD = 4.63, p = .047).Although on average African Ameri-can, Hispanic, and Asian/Pacific Islanderyouth endorsed more internalizing prob-lems than their parents (positive dif-ference scores), Caucasian parents re-ported more internalizing problems thantheir adolescents (negative differencescore). YSR–CBCL externalizing differ-ence scores were significantly associatedwith race/ethnicity, F(3, 570) = 2.94, p =.033), such that Caucasians had lower dif-ference scores than African Americans(Tukey HSD = 3.29, p = .020), Hispanics(Tukey HSD = 2.82, p = .002), andAsian/Pacific Islanders (Tukey HSD =

FIGURE 1. Behavior problems by informant type as a function of race/ethnicity. Note. CA = Caucasian; AA = AfricanAmerican; HA = Hispanic; API = Asian/Pacific Islander; INT = internalizing behavior score; EXT = externalizingbehavior score.


4.68, p = .045). As with internalizing prob-lems, African American, Hispanic, andAsian/Pacific Islander youth endorsedmore externalizing problems than theirparents (positive difference scores); Cau-casian parents reported more externaliz-ing problems than their adolescents (neg-ative difference score).

Adolescent–Teacher Difference Scores.After holding covariates constant, race/ethnicity was significantly associated withYSR–TRF internalizing difference scores,F(3, 565) = 4.59, p = .003, such that Cau-casians received lower difference scoresthan African Americans (Tukey HSD =3.94, p = .003) and Asian/Pacific Islanders(Tukey HSD = 5.38, p = .013). YSR–TRFexternalizing problems were significantlyassociated with race/ethnicity, F(3, 565) =2.94, p = .033, such that Asian/Pacific Is-landers received higher difference scoresthan Caucasians (Tukey HSD = 4.85, p =.048) and African Americans (Tukey HSD = 6.83, p = .010). In terms of ab-solute differences, adolescents on averageself-reported more internalizing and ex-ternalizing problems than were reflectedin the teacher reports.

Parent–Teacher Difference Scores.After holding covariates constant, therewere no significant effects of race/ethnicity or age on CBCL–TRF internal-izing difference scores. However, a sig-nificant effect of race/ethnicity emerged for CBCL–TRF externalizing differencescores, F(3, 565) = 4.86, p = .002, suchthat African American youth receivedlower difference scores than Caucasians(Tukey HSD = 5.24, p < .001) and Asian/Pacific Islanders (Tukey HSD = 5.64, p =.043). In general, parents reported moreproblems than did teachers. However,these differences were less pronouncedfor African American adolescents than forCaucasians and Asian/Pacific Islanders.

Racial/Ethnic Differences inCross-Informant q Correlations

Figure 3 displays the estimated mean q correlations (adjusted for covariates) forthe three dyads, by race/ethnicity for eachproblem type. The preliminary multivari-ate test indicated that race/ethnicity, F(18,1710) = 2.69, p < .001; gender, F(6,568) = 3.94, p = .001; age, F(6, 568) =5.09, p < .001; ADM sector involvement,

F(6, 568) = 2.19, p = .043; and the inter-action between gender and race/ethnicity,F(18, 1710) = 3.495, p < .001, were sig-nificantly associated with the multivariateset of inter-informant q correlations, butfamily income was not. Follow-up uni-variate ANCOVA results for the effects ofrace/ethnicity on each dyad broadband q correlation are presented in the follow-ing sections.

Adolescent–Parent q Correlations.YSR–CBCL internalizing q correlationswere not associated with race/ethnicityafter controlling for age, family income,and ADM sector affiliation. In contrast,YSR–CBCL externalizing q correlationswere associated with race/ethnicity, F(3,576) = 4.12, p = .007, such that Caucasianshad higher q correlations than Hispanics(Tukey HSD = .068, p = .006) and Asian/Pacific Islanders (Tukey HSD = .101, p =.022). The interaction of race/ethnicityand gender was not significantly asso-ciated with YSR–CBCL externalizing q correlations.

Adolescent–Teacher q Correlations.After taking gender, age,ADM sector, and

FIGURE 2. Inter-informant difference scores as a function of race/ethnicity. Note. CA = Caucasian; AA = AfricanAmerican; HA = Hispanic; API = Asian/Pacific Islander; INT = internalizing behavior score; EXT = externalizingbehavior score.


family income into account, there was a significant effect of race/ethnicity onYSR–TRF internalizing q correlations,F(3, 573) = 2.84, p = .037; however, thiseffect was moderated by gender, F(3,573) = 3.51, p = .015. Figure 4 illustratesthe nature of this interaction. Post hoc uni-variate ANCOVAs run separately for girlsand boys indicated that the effect ofrace/ethnicity on YSR–TRF internalizingq correlations was significant for girls,F(1, 183) = 2.92, p = .035, but not for boys.For girls, African Americans had higherYSR–TRF internalizing q correlationsthan Caucasians (Tukey HSD = .204, p =.023). After controlling for covariates,there was also a marginal effect of race/ethnicity on YSR–TRF externalizing qcorrelations, F(3, 573) = 2.60, p = .051;however, this effect was moderated bygender, F(3, 573) = 5.02, p = .002. Fig-ure 4 illustrates the nature of this interac-tion. As was the case with YSR–TRFinternalizing q correlations, post hocanalyses indicated that the effect of race/ethnicity on YSR–TRF internalizing q cor-relations was significant for girls, F(1,183) = 5.83, p = .0.01, but not for boys.For girls, African Americans had higher

YSR–TRF externalizing q correlationsthan Caucasians (Tukey HSD = .100, p =.031) and Hispanics (Tukey HSD = .159,p < .001).

Parent–Teacher q Correlations. Therewas also a significant effect of race/ethnicity on CBCL–TRF internalizing q correlations, F(3, 576) = 5.22, p = .0.01,such that African Americans had lower q correlations than Caucasians (Tukey HSD = .256, p = .007) and Hispanics(Tukey HSD = .234, p = .026) andAsian/Pacific Islanders had lower q cor-relations than Caucasians (Tukey HSD =.305, p = .043). These results held aftercontrolling for age, family income, ADMsector, and gender. The interaction ofrace/ethnicity and gender also had a sig-nificant effect on CBCL–TRF externaliz-ing q correlations, F(3, 573) = 2.71, p =.044. Figure 4 illustrates the nature of thisinteraction. Post hoc univariate ANCOVAsrun separately for the four racial/ethnicgroups indicated that the effect of genderon CBCL–TRF externalizing q correla-tions was significant for Asian/PacificIslanders, F(1, 41) = 4.48, p = .046, butnot for Caucasians, Hispanics, or Afri-

can Americans. Among Asian/Pacific Is-landers, boys had higher CBCL–TRF ex-ternalizing q correlations than girls.

Table 1 presents an overview of the main findings regarding racial/ethnicdifferences in informant reports, inter-informant discrepancies, and inter-informant agreement.

DISCUSSION

The results of this study indicate that re-liance on single informant reports canyield contradictory conclusions regardingracial/ethnic differences in adolescent psy-chopathology. According to parent re-ports, Caucasian adolescents have moreinternalizing and externalizing problemsthan minority youth, suggesting greaterpsychopathology. According to teacherreports, African American youth in gen-eral have fewer internalizing problems,and Asian/Pacific Islander youth havefewer externalizing problems. However,these racial/ethnic differences did notemerge with youth reports. There was lit-tle variability by race/ethnicity in youth-reported problem scores. This pattern offindings suggests that reports of ethnic

FIGURE 3. Inter-informant q correlations as a function of race/ethnicity. Note. CA = Caucasian; AA = African Ameri-can; HA = Hispanic; API = Asian/Pacific Islander; INT = internalizing behavior score; EXT = externalizing behaviorscore.


differences derived from single informantreports must be considered carefully.Studies that report marked racial differ-ences in youth problems using single in-formant data (e.g., Loo & Rapport, 1998;Tolan & Henry, 1996) are valuable in theirown right, but policy implications must bedrawn with caution. Problem rate esti-mates often serve as needs assessmentsfor services (e.g., outreach, prevention)targeting groups at heightened risk. Suchimportant decisions should be based ondata reconciling reports from multiple in-formants.

Our results indicate that race/ethnicitywas frequently associated with patterns of inter-informant discrepancies in ratings of adolescent psychopathology. Patternsof adolescent–parent discrepancy scoresindicate that the reports of ethnic minor-ity parents might not reveal problems thatminority youth self-identify. Whereas pre-vious literature suggests that youth reportfewer behavioral problems than do theirparents, especially with regard to conduct-related problems (e.g., Herjanic & Reich,1997; Hodges, Gordon, & Lennon, 1990;Loeber, Green, Lahey, & Stouthamer-

Loeber, 1989), our data reveal that thispattern holds only for Caucasian dyads. Inthe minority families, youth tended to re-port more externalizing and internalizingproblems than did their parents. Althoughthis general pattern was not predicted,there are competing interpretations forthese racial/ethnic differences in parent–youth discrepancies. First, Caucasian par-ents may be more sensitive reporters ofyouth maladjustment than minority par-ents. This may suggest racial/ethnic dif-ferences in parental monitoring (Bird etal., 2001) or racial/ethnic differences infamiliarity with Western conceptions ofchild mental health. Caucasian parentsmay have more exposure to educationabout psychopathology, and may thus bemore vigilant of behavior problems thanminority parents (Li, Su, Townes, & Var-ney, 1989). It is also possible that minor-ity parents shouldering the burdens ofmigration or discrimination may be lesssensitive to their children’s distress andthus less apt to notice symptoms (Cauceet al., 2002). Second, Caucasian parentsmay be prone to overpathologizing oroverreporting behavior problems in their

adolescent children. Compared to otherracial/ethnic groups, Caucasian parentsmay hold overly negative perceptions oftheir adolescents, perhaps reflecting moreconflicted parent–child relations or hy-pervigilance regarding youth conduct.However, the extant literature currentlyoffers less support for this interpretation.In either case, these findings have impli-cations for service delivery. Since parentsoften initiate help-seeking, racial/ethnicdifferences in parental problem identifi-cation may contribute to higher rates ofunmet mental health need among minor-ity youth (Shaffer et al., 1996).

Examination of discrepancies betweenthe ratings of teachers and those of otherinformants generally supported our firsttwo hypotheses. Compared to ratings pro-vided for Caucasians, teachers reportedfewer internalizing problems when as-sessing African American and Asian/Pacific Islander adolescents. Teachersmay not be primed to expect these types of adjustment difficulties in AfricanAmerican or Asian/Pacific Islander stu-dents. Teachers may attribute observedbehaviors to racial stereotypes (e.g.,Asian

FIGURE 4. Inter-informant q correlations as a function of race/ethnicity and gender. Note. CA = Caucasian; AA =African American; HA = Hispanic; API = Asian/Pacific Islander; INT = internalizing behavior score; EXT = external-izing behavior score.


Americans are quiet because they are def-erent) and not to emotional problems.Teachers may be least likely to reportproblems that are counter to prevailingracial stereotypes. This is consistent withour finding that youth–teacher discrepan-cies reflect an underidentification of ex-ternalizing problems in Asian/Pacific Is-lander students, relative to Caucasian andAfrican American youth. Alternately, it ispossible that these less visible problemsmanifest themselves differently amongminority youth or that minority youth donot reveal signs of these problems atschool.

Teacher expectancies about base ratesof problems in ethnic groups also ap-peared to be related to elevated reports ofexternalizing problems among AfricanAmericans, as we hypothesized. Com-pared to parents, teachers reported moreexternalizing problems for African Amer-icans. These results are consistent withprevious studies of inter-informant dis-crepancies, suggesting that teachers aremore likely than parents to identify exter-nalizing or disruptive problems in African

American youth (Fabrega et al., 1996;Gottlieb et al., 1991; Youngstrom et al.,2000; Zimmerman et al., 1995). This ro-bust finding suggests consistent differ-ences in the perspectives of the two keyinformants who often function as gate-keepers to treatment. Concerns about pos-sible teacher bias are further supported byevidence that teachers of African Ameri-can children report more behavior prob-lems than do trained independent ob-servers (Puig et al., 1999). At the sametime,African American parents may viewchild behavior through a lens shaped bycultural parenting beliefs and values.African American parents tend to makedifferent judgments about the seriousnessand prognosis of child externalizing be-havior problems than teachers and clini-cians (Lambert, Puig, Lyubansky, Rowan,& Winfrey, 2001). Yet another compet-ing explanation for teacher–parent dis-crepancies in ratings of African Americanyouth is that there may be more cross-situational variability in the behavior ofAfrican American youth in home andschool settings.

Whatever the cause of these inter-informant discrepancies, African Ameri-can parents may not see their children’sbehavior as problematic, but their childrenmay nevertheless be referred to special ed-ucation services (Gottlieb et al., 1991).Higher rates of teacher identification ofexternalizing problems among AfricanAmerican youth may partially account fortheir overrepresentation in special educa-tion services for emotional disturbance(McCabe et al., 1999; Oswald et al.,1999), when disruptive disorders are themost prevalent problem type (Garland et al., 2001).

Our results suggest that race/ethnicityhas distinct influences on differences inlevels of problems reported and interrateragreement about symptom patterns. Therewere no significant effects of race/ethnicity on parent–youth agreementabout internalizing symptom profiles.However, with regard to parent–youthagreement about externalizing problems,we found support for our hypothesis thatthere would be lower agreement aboutspecific behavior problems among the

TABLE 1Summary of Significant Racial/Ethnic Differences in Informant

Reports, Discrepancies, and Agreement

Informant/Dyad Internalizing behavior problems Externalizing behavior problems

Single informant reportsAdolescent report No differences No differencesParent report CA > AA CA > AA, API, HATeacher report CA > AA AA > CA > API

Cross-informant discrepanciesAdolescent–parent Lower discrepancies for CAs than for Lower discrepancies for CAs than for

minorities (API, AA, HA) minorities (API, AA, HA)Adolescent–teacher Lower discrepancies for CAs than AAs and APIs Lower discrepancies for CAs and AAs than

for APIsParent–teacher No differences Lower discrepancies for CAs and APIs than

for AAs

Cross-informant q correlationsAdolescent–parent No differences Agreement higher for CAs than for HAs and

APIsAdolescent–teacher Agreement higher for AA girls than for CA girls Agreement higher for AA girls than for CA

and HA girlsParent–teacher Agreement higher for CAs and HAs than for No differences

APIs and AAs

Note. CA = Caucasian; AA = African American; API = Asian/Pacific Islander; HA = Hispanic.


two immigrant family groups (Asian/Pacific Islander and Hispanic) than amongthe Caucasian dyads. Youth in these fam-ilies may differ from their parents in theirlevels of acculturation and thus view theirown behaviors in a different culturalcontext than their parents, resulting indecreased agreement about behaviors thatconstitute problems. Asian/Pacific Is-lander and Hispanic parents may haveseen certain behaviors as oppositional ordisruptive, while the adolescents viewedthese behaviors as normative modes of ex-pression. Additionally, in families with in-tergenerational cultural rifts, youth mayhave reported behavior problems that theywere successful in hiding from theirparents. These intergenerational culturalgaps may have important implications forthe assessment and treatment of behaviorproblems in immigrant families.

In this study, racial/ethnic effects onyouth–teacher agreement were moder-ated by gender. Consistent with findingsof Youngstrom et al.’s (2000) study ofmale adolescents, our results suggest thatfor male students, there were no racial/ethnic differences in youth–teacher agree-ment about internalizing or externalizingproblems. However, there were racial/ethnic differences for girls:Youth–teacheragreement was superior for African Amer-ican girls than for girls from other racial/ethnic backgrounds. This finding is inter-esting and perhaps suggests that concernsabout teacher bias in assessing AfricanAmerican youth behavior problems maybe more restricted to African Americanboys. However, this result was not previ-ously hypothesized and will require fur-ther investigation and replication.

Finally, our results regarding parent–teacher agreement on symptom profileswere surprising. Our data on parent–teacher agreement on internalizing prob-lems ranged from no agreement (q corre-lations close to zero) for Caucasians andHispanics to active disagreement (nega-tive q correlations) for African Americansand Asian/Pacific Islanders. Our results ina service-using sample reflect much moredisagreement than previous studies ofyouth at arguably lower risk (e.g.,Young-strom et al., 2000). It is perhaps not sur-prising that our adult informants in the

home and school settings provided ratingsof emotional problems that are difficult toobserve, yielding unrelated symptom pro-files. However, it is more puzzling to ex-plain sets of ratings that are extremelydivergent. It may be that teachers andAfrican American and Asian/Pacific Is-lander parents hold opposing definitionsof these symptoms, causing perceptionsof these behaviors to be culturallyembedded. Parent–teacher agreement onexternalizing problems was higher andwithin the range reported in previous stud-ies. We found a significant interaction be-tween race/ethnicity and gender, such thatagreement about externalizing problemswas better for Asian/Pacific Islander boysthan Asian/Pacific Islander girls. Thismay be related to both teacher expec-tancies and parental cultural distressthresholds. Based on prevailing stereo-types, teachers may be least likely to lookfor conduct-related problems in Asian/Pacific Islander girls, and Asian/PacificIslander parents may have heightenedcultural-related concerns about conductproblems in their daughters. Again, thesefindings that were not hypothesized awaitreplication and further study before firmconclusions can be drawn.

In summary, race/ethnicity has mul-tiple apparent associations with inter-informant agreement and discrepancies inratings of adolescent psychopathology.These findings have multiple implicationsand potential explanations that can beclassified into three categories: culturaldifferences in adult distress thresholds,racial/ethnic bias among teachers, andcultural differences in symptom expres-sion. A limitation of the current study liesin the inability to test these competinghypotheses. Clarification of the deter-minants of ethnic differences in inter-informant discrepancies and agreement inratings of youth problems awaits future re-search. A second limitation of the studyarises from the composition of our sam-ple. Although the vast majority of our par-ent informants were biological parentsand closely related caregivers, our high-risk sample included some youth in out-of-home placements, and the accuracy ofparent reports may be limited in thesecircumstances.

Notwithstanding these study limita-tions, there are some immediate implica-tions of our results. First, epidemiologicstudies reporting racial/ethnic differencesin youth psychopathology may be mis-leading when based on single informantreports. Second, youth in distress may notbe finding their way to mental health ser-vices if parents are relied upon to triggerhelp-seeking, and this may be especiallytrue for minority youth. Implementationof screening protocols and school-basedservices may provide a safety net for thesevulnerable and underserved youth. Third,our findings suggest that teachers and par-ents from certain minority groups mayneed to reconcile differences in their as-sessments of adolescents in order to arriveat mutually agreeable goals and treat-ments in Individualized Education Pro-grams for special education services.

In particular, our results seem espe-cially relevant to national concerns aboutthe overrepresentation of African Ameri-can youth in special education and juve-nile justice. In the last 20 years, there hasbeen a growing amount of disproportion-ality in the representation of AfricanAmerican youth in placements for emo-tional disturbance (Losen & Orfield, 2002;National Research Council, 2002). Ofparticular concern is the evidence thatthese services may not be meeting theneeds of African American youth. Post-graduation outcomes for this group arestrikingly inferior, with high rates ofsubsequent involvement in the adult orjuvenile correctional systems (Oswald,Coutinho, & Best, 2002). Furthermore,African American youth in schools aremore likely to receive harsh disciplinarypenalties, including suspension and ex-pulsion for behavioral offenses, than areCaucasian youth (Skiba, Michael, Nardo,& Peterson, 2002). Our findings suggestthat these educational decisions are oftensupported by teacher reports of behavioralproblems that are not necessarily corrob-orated by youth self-reports and parentimpressions. For this particularly vulner-able group, it is imperative that alternativesupport services be considered beforeplacement or removal of the youth. Osher,Woodruff, and Sims (2002) proposed aspectrum of more responsive services, in-

cluding universal prevention, targetedearly interventions, and intensive individ-ualized interventions within mainstreameducational placements.

About the Authors

ANNA S. LAU, PhD, is an assistant professorof psychology at the University of California–Los Angeles. Her research focuses on ethno-cultural issues in child psychopathology, childmaltreatment, and delivery of children’s men-tal heatlh services and interventions. ANN F.GARLAND, PhD, is an associate professor ofpsychiatry at the University of California–SanDiego and an associate director of the Childand Adolescent Services Research Center atSan Diego Children’s Hospital and HealthCenter. Her primary research interests includeoutcome assessment and evidence-based prac-tice in children’s mental health services. MAYYEH, PhD, is an assistant professor of psy-chology in the SDSU/UCSD Joint DoctoralProgram in Clinical Psychology at San DiegoState University, research scientist at the Child and Adolescent Services Research Cen-ter at San Diego Children’s Hospital andHealth Center, and adjunct assistant professorof psychiatry at the University of California–San Diego. Her primary research interests arein cultural issues in mental health service de-livery to children. KRISTEN M. MCCABE,PhD, is a research scientist at the Child andAdolescent Services Research Center at SanDiego Children’s Hospital and Health Center,assistant professor of psychology at the Uni-versity of San Diego, and adjunct assistantprofessor of psychiatry at the University ofCalifornia–San Diego. Her current profes-sional work and interests include studying cul-tural factors that affect mental health serviceaccess and effectiveness for minority childrenand adolescents and the development of cul-turally responsive treatment modalities foryoung children. PATRICIA A. WOOD,MPH, is a senior data analyst at the Child andAdolescent Services Research Center at SanDiego Children’s Hospital and Health Center.Her primary research interests include accessto health services for underserved groups.RICHARD L. HOUGH, PhD, is emeritusprofessor of sociology at San Diego State Uni-versity, research professor of psychiatry andfamily and community medicine at the Uni-versity of New Mexico, adjunct associate pro-fessor of psychiatry at the University ofCalifornia–San Diego, co-director of the Childand Adolescent Services Research Center, andprincipal investigator on the Patterns of Care

study. His research interests include pathwaysinto and through public sectors of care foryouth. Address: Anna S. Lau, PhD, Universityof California, Los Angeles, Department ofPsychology, 1285 Franz Hall, Box 951563,Los Angeles, CA 90095-1563; e-mail: [email protected]

Authors’ Notes

1. This study was funded by National Insti-tute of Mental Health (NIMH) Grant U01MH55282 (PI: Dr. Richard L. Hough). Thepreparation of this article was supported byResearch Scientist Development AwardsK01 MH01767, K01 MH01924, and K01MH01544.

2. We gratefully acknowledge the collabora-tion of the San Diego County Health andHuman Services Agency, San DiegoCounty Probation Department, and the SanDiego County Office of Education. We arealso indebted to the many families who par-ticipated in this study.

Note

We further compared each constituent AN-COVA model produced by SPSS with a com-parable regression model using STATA Ver-sion 7.0 to take both sampling weights andstratification design effects into account. Thetwo methods of analysis produced highly com-parable results (i.e., same pattern of significantfindings for all race/ethnicity and gender ef-fects). Therefore, we chose to report the orig-inal SPSS MANCOVA results because the pre-liminary MANCOVA controls for Type I errorwith the multiple dependent variables underexamination, and these analyses permitted es-timation of marginal means by gender andrace/ethnicity to aid interpretation of sig-nificant gender by race/ethnicity interactionterms.

References

Achenbach, T. M. (1991a). Integrative guidefor the 1991 CBCL/4-18,YSR and TRF Pro-files. Burlington: University of Vermont,Department of Psychiatry.

Achenbach, T. M. (1991b). Manual for theChild Behavior Checklist/4-18 and 1991Profile. Burlington: University of Vermont,Department of Psychiatry.

Achenbach, T. M. (1991c). Manual for theTeachers Report Form and 1991 Profile.Burlington: University of Vermont, Depart-ment of Psychiatry.

Achenbach, T. M., McConaughy, S. H., &Howell, C. T. (1987). Child/adolescent be-havior and emotional problems: Implica-tions of cross-informant correlations for sit-uational specificity. Psychological Bulletin,101, 213–232.

Bannai, H., & Cohen, D. A. (1985). The pas-sive-methodical image of Asian Pacific Is-lander students in the school system. Soci-ology and Social Research, 70, 79–81.

Bird, H. R., Shaffer, D., Fisher, P., Gould,M. S., Staghezza, B., Chen, J. Y., & Hoven,C. (1993). The Columbia Impairment Scale(CIS): Pilot findings on a measure of globalimpairment for children and adolescents. In-ternational Journal of Methods in Psychi-atric Research, 3, 167–176.

Cauce, A. M., Domenech-Rodriguez, M., Par-adise, M., Cochran, B. N., Shea, J. M., Sreb-nik, D., & Baydar, N. (2002). Cultural andcontextual influences in mental health helpseeking: A focus on ethnic minority youth.Journal of Consulting and Clinical Psy-chology, 70, 44–55.

Chang, D. F., & Sue, S. (2003). The effects ofrace and problem type on teachers’ assess-ments of student behavior. Journal of Con-sulting and Clinical Psychology, 71, 235–242.

Dusek, J. B., & Joseph, G. (1983). The basesof teacher expectancies: A meta-analysis.Journal of Educational Psychology,75, 327–346.

Fabrega, H., Ulrich, R., & Loeber, R. (1996).Adolescent psychopathology as a functionof informant and risk status. Journal of Ner-vous and Mental Disease, 184, 27–34.

Frick, P. J., Silverthorn, P., & Evans, C. (1994).Assessment of childhood anxiety usingstructured interviews: Patterns of agreementamong informants and association with ma-ternal anxiety. Psychological Assessment, 6,372–379.

Garland, A. F., Hough, R., McCabe, K. M.,Yeh, M., Wood, P., & Aarons, G. (2001).Prevalence of psychiatric disorders inyouths across five sectors of care. Journalof the American Academy of Child and Ado-lescent Psychiatry, 40, 409–418.

Gottlieb, J., Gottlieb, B. W., & Trongone, S.(1991). Parent and teacher referrals for psy-choeducational evaluation. The Journal ofSpecial Education, 25, 155–167.

Henry, G. T. (1990). Practical sampling. New-bury Park, CA: Sage.

Herjanic, B., & Reich, W. (1997). Develop-ment of a structured psychiatric interviewfor children: Agreement between child andparent on individual symptoms. Journal ofAbnormal Child Psychology, 25, 21–31.


Hodges, K., Gordon, Y., & Lennon, M. P.(1990). Parent-child agreement on symp-toms assessed via a clinical research inter-view for children: The Child AssessmentSchedule (CAS). Journal of Child Psychol-ogy and Psychiatry, 31, 427–436.

Kolko, D. J., & Kazdin, A. E. (1993). Emo-tional/behavioral problems in clinic andnonclinic children: Correspondence amongchild, parent and teacher reports. Journal ofChild Psychology, Psychiatry and AlliedDisciplines, 34, 991–1006.

Lambert, M. C., Puig, M., Lyubansky, M.,Rowan, G. T., & Winfrey, T. (2001). Adultperspectives on behavior and emotionalproblems in African American children.Journal of Black Psychology, 27, 64–85.

Lethermon, V. R., Williamson, D. A., Moody,S. C., & Wozniak, P. (1986). Racial bias inbehavioral assessment of children’s socialskills. Journal of Psychopathology & Be-havioral Assessment, 8, 329–337.

Li, X., Su, L., Townes, B. D., & Varley, C. K.(1989). Diagnosis of attention deficit disor-der with hyperactivity in Chinese boys.Journal of the American Academy of Child& Adolescent Psychiatry, 28, 497–500.

Loeber, R., Green, S. M., Lahey, B. B., &Stouthamer-Loeber, M. (1989). Optimal in-formants on childhood disruptive behaviors.Development and Psychopathology, 1, 317–337.

Loo, S. K., & Rapport, M. D. (1998). Ethnicvariations in children’s problem behaviors:A cross-sectional, developmental study ofHawaii school children. Journal of ChildPsychology, Psychiatry and Allied Disci-plines, 39, 567–575.

Losen, D. J., & Orfield, G. (Eds.). (2002).Racial inequity in special education. Cam-bridge, MA: Harvard Education Press.

Marsella, A. J., & Kameoka, V. A. (1989).Ethnocultural issues in the assessment ofpsychopathology. In S. Wetzler (Ed.), Mea-suring mental illness in psychometric as-sessment for clinicians (pp. 231–256).Washington, DC: American Psychiatric As-sociation.

Marwit, S. J. (1982). Students’ race, physicalattractiveness and teachers’ judgments fol-lowing transgressions. Psychological Re-ports, 50, 242.

McCabe, K. M., Yeh, M., Hough, R. L.,Landsverk, J., Hurlburt, M. S., Culver, S. W.,& Reynolds, B. (1999). Racial/ethnic repre-sentation across five public sectors of carefor youth. Journal of Emotional and Be-havioral Disorders, 7, 72–82.

National Research Council. (2002). Minoritystudents in special and gifted education.Washington, DC: National Academy Press.

Osher, D., Woodruff, D., & Sims,A. E. (2002).Schools make a difference: The overrepre-sentation of African American youth in spe-cial education and the juvenile justice sys-tem. In D. Losen & G. Orfield (Eds.), Racialinequity in special education (pp. 93–116).Cambridge, MA: Harvard Education Press.

Oswald, D. P., Coutinho, M. J., & Best, A. M.(2002). Community and school predictorsof overrepresentation of minority childrenin special education. In D. Losen & G. Or-field (Eds.), Racial inequity in special edu-cation (pp. 1–13). Cambridge, MA: HarvardEducation Press.

Oswald, D. P., Coutinho, M. J., Best, A. M., &Singh, N. N. (1999). Ethnic representationin special education: The influence ofschool-related economic and demographicvariables. The Journal of Special Education,32, 194–206.

Pigott, R., & Cowen, E. (2000). Teacher race,child race, racial congruence, and ratings ofchild school adjustment. Journal of SchoolPsychology, 38, 177–196.

Prieto, A. G., & Zucker, S. H. (1981). Teacherperception of race as a factor in the place-ment of behaviorally disordered children.Behavioral Disorders, 7(1), 34–38.

Puig, M., Lambert, M. C., Rowan, G. T., Win-frey, T., Lyubansky, M., Hannah, S. D., &Hill, M. F. (1999). Behavioral and emo-tional problems among Jamaican andAfrican American children, ages 6 to 11:Teacher reports versus direct observations.Journal of Emotional and Behavioral Dis-orders, 7, 240–250.

Reid, R. (1995). Assessment of ADHD withculturally different groups: The use of be-havioral rating scales. School PsychologyReview, 24, 537–560.

Schneider, B., & Lee, Y. (1990). A model foracademic success: The school and home en-

vironment of Asian students. Anthropologyand Education Quarterly, 21, 358–377.

Shaffer, D., Fisher, P., Dulcan, M. K., Davies,M., Piacentini, J., Schwab-Stone, M., et al.(1996). The NIMH Diagnostic InterviewSchedule for Children Version 2.3: De-scription, acceptability, prevalence rates,and performance in the MECA study. Jour-nal of the American Academy of Child andAdolescent Psychiatry, 35, 865–877.

Skiba, R. J., Michael, R. S., Nardo, A. C., &Peterson, R. L. (2002). The color of disci-pline: Sources of racial and gender dispro-protionality in school punishment. UrbanReview, 34, 317–342.

Szapocznik, J., & Truss, C. (1978). Intergen-erational sources of conflict in Cuban moth-ers. In M. Montiel (Ed.), Hispanic families(pp. 41–65). Washington, DC: NationalCoalition of Hispanic Mental Health andHuman Services Organization.

Tolan, P. H., & Henry, D. (1996). Patterns ofpsychopathology among urban poor chil-dren: Comorbidity and aggression effects.Journal of Consulting and Clinical Psy-chology, 64, 1094–1099.

Waller, N. G., & Meehl, P. E. (1998). Multi-variate taxometric procedures: Distinguish-ing types from continua (Vol. 9). ThousandOaks, CA: Sage.

Weisz, J. R., McCarty, C. A., Eastman, K. L.,Chaiyasit, W., & Suwanlert, S. (1997). De-velopmental psychopathology and culture:Ten lessons from Thailand. In S. Luthar,J. Burack, D. Cicchetti, & J. Weisz (Eds.),Developmental psychopathology: Perspec-tives on adjustment, risk, and disorder(pp. 568–592). New York: Cambridge Uni-versity Press.

Youngstrom, E., Loeber, R., & Stouthamer-Loeber, M. (2000). Patterns and correlatesof agreement between parent, teacher, andmale adolescent ratings of externalizing andinternalizing problems. Journal of Consult-ing & Clinical Psychology, 68, 1038–1050.

Zimmerman, R. S., Khoury, E., Vega, W., Gil,A., & Warheit, G. (1995). Teacher andparent perceptions of behavior problemsamong a sample of African American, His-panic, and Non-Hispanic White students.American Journal of Community Psychol-ogy, 23, 181–197.


race/ethnicity and inter-informant agreement in …...highly valued, reports from adolescents,...

Documents