Parent Reports of Behavioral and EmotionalProblems among Children in Kenya, Thailand,and the United States

John R. Weisz and Marian Sigman

University of California, Los Angeles

Bahr Weiss

Vanderbilt University

Julie Mosk

University of California, Los Angeles

WEISZ, JOHN R. ; SIGMAN, MARIAN; WEISS, BAHH; and MOSK, JULIE. Parent Reports of Behavioraland Emotional Problems among Children in Kenya, Thailand, and the United States. CHILDDEVELOPMENT, 1993, 64, 98-109. The behavioral and emotional problems children developmay differ from one cultural context to another. We explored this possibility, comparing 11—15-year-old Embu children in Kenya, Thai children, African-American children, and Caucasian-American children. Standardized parent reports on 118 problems revealed 62 significant (p <.01) culture effects. Caucasian-Americans were rated particularly high on undercontrolled prob-lems (e.g., arguing, disobedient at home, cruel to others). Embu children were rated particularlyhigh on overcontrolled problems (e.g., fears, feels guilty, somatic concerns), largely because ofthe numerous somatic problems reported. The findings may relate to the strict emphasis oncompliance and obedience among the Embu, as opposed to the greater independence permittedin the United States. But alternative interpretations are discussed as well, including the effectsof parent sensitivities and Third World living conditions.

A growing body of research supports the Phares, & Grayson, 1990), Asia (Weisz et al.,notion that patterns of child and adolescent 1987); and Europe (Achenbach, Verhulst,psychopathology are infiuenced by the cul- Baron, & Akkerhuis, 1987).ture in which youngsters grow up. Cultur-ally mediated beliefs, values, and traditions In contrast to the changing state of af-and associated child-rearing and socializa- fairs in many regions of the world, we knowtion practices may help shape both the kinds relatively little about child and adolescentof problems youngsters show when dis- problem behavior in Africa. There havetressed (see, e.g., Lambert, Weisz, & Knight, been several relevant studies, a number of1989; Weisz, 1989) and the kinds of prob- these in Kenya; some have dealt with child-lems parents perceive or find distressing rearing and child behavior generally (e.g.,(e.g., Weisz et al., 1988). To explore this pos- Levine & Levine, 1977; Whiting & Whiting,sibility, researchers have begun to develop 1975), whereas others have dealt with spe-a base of data on child and adolescent prob- cific forms of problem behavior—for exam-Iem behavior in various parts of the world, pie, learning disabilities and mental retar-including North America (Achenbach & dation (Dhadphale & Ibrahim, 1984) andEdelbrock, 1981), South America (Monte- symptoms of depression and anxiety (Mitch-negro, 1983), the Caribbean (Lambert et ell & Abbott, 1987). However, to our knowl-al., 1989), Australia (Achenbach, Hensley, edge, none of these studies has surveyed a

The research was supported through grants from NIMH to John Weisz (1 ROl MH 38240)and Thomas Achenbach (1 ROl MH 40305) and through USAID Contract DAN 1309-G-SS-80(Project Director: Doris Calloway). We thank Susan D'Sousa and Nellie Matita for conductingthe interviews with Embu parents, and Pirapong Kitlvejpokawat, Samart Limdhepatip, and Su-wattana Sripuenpol for conducting the Thai interviews. We also thank Thomas Achenbach, Wan-chai Chaiyasit, Mike Espinoza, Sherryl Goodman, Catherine Howell, and Somsong Suwanlertfor sharing data with us and for their help with other aspects of the project. Address reprintrequests to John Weisz, Department of Psychology, Franz Hall, UCLA, 405 Hilgard Avenue, LosAngeles, CA 90024-1563.

[Child Development, 1993, 64, 98-109. © 1993 by the Society for Research in Child Development, Inc.AH rights reserved. 0009-3920/93/6401-0009$01.00]

Weisz et al. 99

broad array of problems, none has used astandard problem assessment procedure,and none has directly compared the reportedproblems of African youngsters to those ofyouth from other regions. These were objec-tives of the present research.

To maximize the theoretical and empiri-cal yield of such research, it is important toselect comparison cultures that are differentin conceptually significant ways and to focuson problem patterns, or syndromes, whichtheory suggests may be related to those cul-tural differences. The syndromes selectedfor study here were the two that are mostfrequently identified in factor analytic re-search with children: overcontrolled or "in-ternalizing" problems (e.g., shyness, fear-fulness, somaticizing), and undercontrolledor "externalizing" problems (e.g., fighting,showing off, hyperactivity). The two syn-dromes have been found in more than 20factor analytic studies (many reviewed byAchenbach & Edelbrock, 1978). Althoughwe do not know a great deal about what pro-vokes the development of over- and under-controlled problems, some evidence (Ach-enbach & Edelbrock, 1978; Hetherington &Martin, 1986) and theory (Weisz, 1989) sug-gest that socialization that stresses self-control, emotional restraint, and strict com-pliance with social rules may make childrenmore likely to develop overcontrolled as op-posed to undercontrolled problems.

The African youngsters of interest in thepresent study were children living in theEmbu District of east-central Kenya (for de-tails on the people and region, see Sigmanet al., 1988; Sigman, Neumann, Jansen, &Bwibo, 1989; see also below). The Embupeople are a part of the large Bantu streamthat populates much of sub-Saharan Africa.Like many African peoples, they reside pri-marily in rural areas and live by cultivatingcrops, herding animals, and sometimesworking at jobs that provide modest income(e.g., driver, teacher). Most infants and tod-dlers accompany their mothers throughoutthe day; most older children are in schoolmuch of the day and at home late afternoonsand evenings. Given the literature on the or-igin of overcontrolled problems (above), it isintriguing to note that when Embu parentswere asked (in an unpublished portion ofSigman et al., 1989) what they wanted mostfor their children, 92% replied with obedi-ence, politeness, or both. As for African chil-dren more generally, Munroe and Munroe(1977) have noted that, "In sub-Saharan Af-rica, socialization almost universally empha-sizes strong pressures [on children] toward

compliance" (p. 146). Decades ago, in Ken-yatta's (1938) classic study of the Embu'sneighboring tribe, the Gikuyu, he noted,"Growing boys and girls learn that they haveone thing to learn which sums up all the oth-ers, and that is the manners and deportmentproper to their station in the community.They see that their happiness in the home-stead, their popularity with their playmates,their present comforts, and their future pros-pects depend on knowing their place, givingrespect and obedience where it is due"(p. 106).

The groups selected for comparisonwith the Embu youth were youngsters fromThailand and the United States (African-American and Caucasian groups consideredseparately) who, like the Embu, lived out-side cities and suburbs. Thai youngsters of-fer a useful comparison group because theirrearing conditions, and the socializationgoals that guide their rearing, appear to besimilaur in several respects to those of Ke-nyan youth. In Thai society, children are so-cialized to be obedient and polite, to avoidoffending or distressing others (particularlythose who are older), and to show emotionalrestraint and self-control (see Boesch, 1977;Suvannathat, 1979; Suwanlert, 1974; Weisz,1989). By contrast, American youngsters ap-pear to be given considerably greater lat-itude, with independence and emotionalexpression not only tolerated but often en-couraged and rewarded (see Weisz, 1989).

In previous epidemiologic research re-lying on both parent reports and teacher re-ports (Weisz et al , 1987,1992) Thai childrenand adolescents have consistently beenrated higher than American children onovercontrolled problems. Some theoristshave speculated that Thai children may be-come more overcontrolled or internalizingin part because of socialization that stressesself-control, social inhibition, and compli-ance with social norms (see Boesch, 1977;Sangsingkeo, 1969; Weisz, 1989). Followingthis reasoning, one might expect from thebrief child-rearing overview given abovethat parents of Embu children, like parentsof Thai children, would report higher levelsof overcontrolled behavior than the twoAmerican groups, because the Embu peo-ple, like Thais, emphasize self-control andsocial compliance. However, it is also possi-ble that parents would be sensitized to prob-lems of undercontrol in a society wherecompliance and obedience are important (al-though the Thai data do not appear tosupport this hypothesis). Furthermore, theEmbu society, like most, must certainly con-

100 Child Development

vey a variety of messages to its youth (see,e.g., Saberwal's [1970] report on Embu prac-tices that appear to encourage assertivenessand aggression), and there is clearly not suf-ficient evidence for a firm prediction at thispoint.

As the preceding paragraph implies, re-liance on parent reports of child problemsraises a number of intriguing conceptual is-sues. One issue is the nature of a "childproblem." The concept of a child problem,like the concept of child psychopathologygenerally, must inevitably involve two com-ponents: actual child behavior and the per-ception of that behavior through the lensof some observer—a parent, a teacher, atrained observer, or the child him- or herself.Indeed, one could argue that "actual behav-ior" and "perceived behavior" are inextrica-bly linked in reports by any informant. A va-riety of factors may infiuence whether aparticular observer reports a particular prob-lem for a child—factors such as whether thebehavior is unusual or otherwise salient inits cultural and social context, and evenwhether the observer is willing to report theproblem to an interviewer. In the long term,there is value in sampling reports from mul-tiple informants. However, assessing parentreports may be a particularly useful firststep, because (1) parents have a particularlycomprehensive observational base (i.e.,more exposure to their child across more set-tings than, say, teachers or trained observ-ers), (2) parents may have somewhat betterdescriptive language skills than their chil-dren, and (3) parent reports reflect the com-bination of actual child behavior and cultur-ally influenced perception that is theessence of "child problems" in the firstplace.

Accordingly, we used parent-reportmethodology to help extend the base of in-formation on child problems to an Africansample. We obtained parent reports on theprevalence of 118 specific behavior prob-lems among Embu children; these data werecompared to parallel parent reports for Thai,African-American, and Caucasian-Americanyouth. The groups were compared not onlyon the specific problems but also on totalproblem scores and on the two most com-monly identified syndromes of problembehavior—overcontrolled and undercon-trolled. We explored problem prevalence asa function of culture, gender, and the inter-action of culture and gender.

MethodSubjects and Research Design

The total sample numbered 308, with90 Embu youth, 96 Thais, 32 African-Americans, and 90 Caucasian-Americans. Allchildren lived in rural or semirural areas.Each group had equal numbers of boys andgirls, the same age range (11—15), and simi-lar age means (13.0, 13.1, 12.9, and 13.0, re-spectively). A 4 (culture) X 2 (gender) AN-OVA on age yielded no significant effects ofculture, gender, or their interaction (all p's> .50). The sample formed a 4 (culturegroup) X 2 (gender) design with propor-tional cell N's throughout.

We were unable to find SES classifica-tion schemes for Kenya or Thailand whichhad either validity information or suffi-ciently detailed coding procedures to permitthem to be applied to our data. So, instead,we applied Hollingshead's (1975) SES sys-tem to parent occupation data in each coun-try. In the economies of Thailand and Ke-nya, some of the logic of the Hollingsheadsystem does apply; for example, in all fourcountries, teachers, clerks, and small busi-ness operators do tend to have higher levelsof education and income than subsistencefarmers and manual laborers. Nonetheless,the SES data for the Embu and Thai samplesshould certainly be interpreted with cautionbecause they involve use of an Americansystem in cultures for which it was not de-signed. To provide for at least a rough simi-larity on parent occupational level, we se-lected samples with the constraint that thefour groups not differ on this SES measure.Mean ratings on Hollingshead's 9-pointscale (9 = highest SES) were 3.42, 3.62,3.38, and 3.83 for the Embu, Thai, African-American, and Caucasian samples, respec-tively. The overall culture effect was nonsig-nificant (p > .5).

Study Areas and Sampling ProceduresEmbu sample.—We sampled three loca-

tions in the Embu District, in central Kenya,about 120 miles northeast of Nairobi. Thedistrict rests on the slopes of Mt. Kenya,where the Embu people live through a com-bination of subsistence farming (corn, beans,livestock), modest small businesses (e.g.,cash crops, such as coffee), and private andpublic employment (e.g., manual laborer,driver, teacher). The children were origi-nally selected because they were between 7and 9 years of age at the time of a previousstudy focused on schoolchildren (reported inSigman et al., 1989).

Weisz et al. 101

Thai sample.—Because the Embuyouth resided outside the boundaries of anycity or suburb, we used comparison samplesof Thai and American youth who did not livein cities or suburbs. Thai youngsters weresampled from districts in the central, north,nordieast, and southern peninsula areas ofThailand. As with the Embu sample, theThai parents live through a combination ofsubsistence farming, small businesses (e.g.,cash crops such as bananas), and public andprivate employment. The children were ran-domly sampled through the use of schooland province population directories (seeWeisz et al., 1987, 1989 for further details).

U.S. samples.—^Both the black andwhite U.S. samples were drawn from numer-ous regions throughout the United States,'selected for representativeness by the In-stitute of Survey Research at Temple Uni-versity (procedures described in Stanger,McConaughy, & Achenbach, in press). Oursample, a subset of the Stanger et al. sample,included black and white subjects, selectedrandomly with the constraints that they notdiffer from one another or from the Thai orEmbu samples in parental occupational lev-els (see SES data, above) and that they notreside in cities or suburbs. The range of par-ent occupations included farming as well asother private and public employment.

Problem Checklist Measure andInterview Procedure

In the United States, the problem reportmeasure was the Child Behavior Checklist(CBCL; Achenbach, 1991). It includes ques-tions about child and parent demographics,child competence in school and elsewhere,and, a list of 118 specific problems (e.g., "dis-obedient at home," "shy or timid"). Parentsindicate the degree to which their childshows each problem by ratings of 0 = nottrue of the child, 1 = somewhat or some-times true, or 2 = very true or often true.The CBCL is very widely used and wellsupported through extensive psychomet-ric data (see Achenbach, 1991). Principalcomponents analyses of CBCL data haverevealed overcontrolled ("internalizing")and undercontrolled ("externalizing") syn-dromes, as described above.

The Thai and Kenyan parents com-pleted checklist measures that included thesame problem items as the CBCL, translatedinto Thai and Kiembu, respectively. Thesetwo checklists were designed to be suffi-ciently similar in format and content to theCBCL to permit cross-cultural comparisons,but also sufficiently sensitive to Thai andEmbu cultures to detect patterns of particu-lar importance in those cultures. The first118 problem items listed in both checklistsare the same (and in the same order) as the118 CBCL problem items, and the responseformat is the same as in the CBCL (i.e., the0-1-2 rating scale). The only exception is thatCBCL item no. 105 ("uses drugs and alco-hol") was divided into two items, "usesdrugs" and "uses alcohol," in the Thai andEmbu checklists; for the present study, how-ever, we combined these two into one item(giving to the composite the higher score ofthe two items), thus forming in effect a sin-gle problem item, as in the CBCL. Addi-tional problem items were added to reflectchild problems of interest in Thailand andKenya, respectively, but the present analy-ses focused only on the initial 118 problemitems, which were the same in all threechecklists.

Conversion of the CBCL items and in-structions into Thai and Kiembu was accom-plished (following Brislin, 1970; Draguns,1982; Wagatsuma, 1977) through successivewaves of translation and backtranslation(three waves in Thailand; two waves inKenya). Each wave involved hired transla-tors and psychologists, aiming for linguisticequivalence and for simplicity of expression(for further details, see Weisz et al., 1987).

All data were collected via individualinterviews with subjects' parents or guard-ians. In each case, the interviewer readaloud the standard problem checklist in theparents' language and recorded the parent'sanswers.

Results

We assessed group differences in (a) to-tal problem scores (i.e., sum of all 1 and 2ratings across all 118 problems plus "other

' The U.S. data come from a large survey of child problems conducted by Stanger, McCo-naughy, and Achenbach (in press). The children used in that survey were sampled from 100locations in the United States identified by the Temple University Institute of Survey Researchas demographically representative of the country as a whole. Identification of areas in which thechildren resided took place 3 years prior to the Stanger et al. survey; thus it is possible that someof the children in our sample no longer lived in rural or semirural areas by the time (3 yearslater) when the problem reports used in the present study were collected.

J

IIOeaaZO

qV

o

ICM

G-OA

m

OCO

mA

O

siPQ

A

O

S3OA

m

MPQr

MM"fM^mI pq I

KH^^M^PMimi

M oa'as"

^ ^ f Hi3=1

I c q t - CO CD •

fH M M H !A A A A A A A A A

A A A A A A A A A^— L^ pn pQ PQ [̂- cG CQ PQ

A A A A A A A A A

^ ^ ""I ̂ IO

P Q H H M H

A A A A A

M M M H M

A A A A A

H PQ ffl ̂ PQ

A A A A A

I <M ~; pCO ^ t - <M.—vO5 00PodE^ 22-52.

H M M W M H M M H PQ M M ̂ ffl M H HA A A I I A A A A A A A A A A A A APQ pq H M H M PQ CQ M M PQ PQ pq H H PQ fflA A A A A A A A A A A A A A A A A

A A A A A A A A A A A A A A AA A

ljllj« .^ .?

3 <D

60 ? Oi « >ij 5̂ JH

8 I -f;

PQ

fa

OO

ftS5JC«ONa3'"'i''

P fa (^

'lnoidrHcMcoincDcDi-*M<ininioininicin

I 00CQ PP

102

OAPQ

CD

sioA

m

i ni-H

oAm

oAPQ

oApq

OAPQ

M" g ^ - ^

cq^MH^Mpq^^Hg^MHHg ^ pq pq ̂ fe! M'fcl M"MM Hffl^PQ H M ^ M ^ g H M ^ M ^ , E.fcl,^^SS^£ fciSs

o"•:;^S'oo',.^'-^ ,-^--^^co'co',-^^o"oo'c5',-^co' ^oTcJ^-^iPrco"..-^^^ ,^.-^.-^00o j S 05 r - i ' ' ^ P ' I ' 1 1 0 CO CD <° CO O5 in T|! o? d r-i CM d <^ 10 CO > > ' - ! c q ' D <= r.^

H ip^H HM ̂ PQ MM PQM PQM M M pq M M M M ^ M M M P Q M M M PQ M

A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A

^;g:^CQpqHui^^HH^HHpq^^H pq^^PQHHMH pqHHPQA A A A A A A A A A A A A A A A A A A A A A A A A A A A A A

AAA A

M MM M H H M ̂ M H M H H M PQ ̂ H

) T3 MJCDCDlr^OOrHCMCO-*!)inininiocDCDcDCD<

> rH CM • * G5 CM) O O O O I—I

O O O P IOOpqO O O O

o ^

T! 2c a

Ill2 H

II8

oV

IsTl 0)

I

A § 2

Bl-

= A '

j 3 '- (U aj

goa3 wS> OS

103

104 Child Development

pbysical problems" and "additional prob-lems"), (il) ratings on each of the 118 specificproblems, and (c) composite scores for over-and undercontrolled problems (see below).Given the large sample and resultant statisti-cal power, we set alpha at .01 rather than.05. We also assessed the magnitude of allsignificant effects by assessing percent vari-ance accounted for; these figures were inter-preted using Cohen's (1988) criteria: effectsare small if they account for l-<5.9% of thevariance, medium if 5.9%—13.8%, and largef

Total Problem ScoresThe 4 x 2 (culture x gender) ANOVA

on total problem scores revealed only onesignificant effect, a main effect of culture,F(3, 300) = 7.23, p = .0001; the effect ac-counted for 7.2% of the variance and wasthus medium by Cohen's (1988) criteria. To-tal problem means were 26.57 for Embus,19.73 for Thais, 22.19 for U.S. blacks, and29.57 for U.S. whites. Pairwise comparisonsrevealed that the Embu and U.S. whitescores were significantly higher than theThai scores (both p's < .01), but that no othergroup differences were significant.

Individual ProblemsThe 4 x 2 (culture x gender) ANOVAs

on the 118 specific problems revealed maineffects of culture on 62 (i.e., more than half)of the problems. The effects are shown inTable 1, together \vith the percent varianceaccounted for by eacb effect and the resultsof pairwise group comparisons. Of the 62 ef-fects, 19 were large, 20 were medium, and23 were small, by Cohen's standards. Embuyoungsters had notably higher scores thanthe other groups on a number of overcon-trolled problems. Quite striking amongthese group differences were the very largeculture effects on problems involving so-matic concerns: no. 56a "aches and pains"(75% of variance), no. 56b "headaches"(74%), no. 56c "nausea, feels sick" (85%), no.56d "problems with eyes" (80%), no. 56g"vomiting" (50%), and no. 76 "sleeps lessthan most children" (52%). These problemsall showed much higher prevalence inEmbu cbildren than in all three of the com-parison groups (all p's < .0001). We followup on these findings below.

Table 1 also shows that there were eightgender effects, six involving higher problemratings for boys than girls. Neither of the twoproblems on wbicb girls scored higher (i.e.,no. 29 "fears," no. 44 "bites nails") were un-dercontrolled (for girls). By contrast, three ofthe six problems on which boys scoredhigher were undercontrolled problems. Thegender effects tended to be more modest inmagnitude tban the culture effects reportedabove; all gender effects were small, by Co-hen's (1988) standards.

Finally, the table shows that only threeof the 118 problems were associated with agender x culture interaction, and all threeinteractions were small in magnitude. Be-cause this number of significant effectsapproximates cbance expectancy, the inter-actions were not analyzed further or inter-preted.

Composite Overcontrolled andUndercontrolled Scores

To compare groups on over- and under-controlled problems, we calculated twocomposite scores for each child, following aprocedure used in previous research (e.g.,Weisz et al., 1987). To calculate each child'sovercontrolled score, we computed thechild's mean rating across all problems thatfit the empirically derived overcontrolledsyndrome for that child's age and gendergroup in principal components analyses byAchenbach and Edelbrock (1983).^ Thus, thescale for these composite scores was thesame as the scale for individual items, 0 to2. Using a parallel procedure, we calculatedan undercontrolled composite score. Thenwe included these two composite scoresin a 4 X 2 X 2 (culture X gender x prob-lem type) repeated-measures ANOVA, withproblem type (i.e., over- vs. undercontrolledscore) as a within-subjects factor.

This analysis produced a significant!main effect for problem type, F(l, 300) =13.96, p = .0001 (accounting for 1.0% of tbevariance), plus significant interactions ofproblem type x gender, F(l, 300) = 8.13, p<.OO1 (accounting for <1% of tbe variance)and problem type X culture, F(l, 300) =65.63, p < .0001 (a medium effect, 9.8% ofthe variance). The main effect for problem

^ As noted elsewhere (e.g., Weisz et al., 1987), a limitation of this procedure is that it isbased on principal components analyses of data from American samples. This seems appropriate,in that such analyses have not been done with Kenyan or Thai children; moreover, it is usefulto have one set of results on which to base item classification, and the U.S. analyses were carefullydone and with very large samples. However, it will be useful in the long run to apply principalcomponents analyses to child problem data from Thailand, Kenya, and other countries as well,to assess degree of correspondence in the composition of child clinical syndromes across cultures.

Weisz et al. 105

Overcontrolled Undercontrolled

I Embu rhai nS3u.S. Black I U.S. White

FIG. 1.—Overcontrolled and undercontrolled problem means for Embu, Thai, U.S. black, and U.S.white children.

type refiected somewhat higher scores forovercontrolled problems (mean: 0.27) thanfor undercontrolled (0.23). The problemtype X gender interaction refiected tbe factthat boys and girls differed in undercon-trolled problems (means: 0.26 and 0.19; p< .0001) but not in overcontrolled problems(boilh means: 0.27).

Underlying the culture X problem typeinteraction were significant effects of prob-lem type for U.S. whites, F(l, 89) = 15.79,p < .0001, and for Embu youth, F(l, 89) =214.71, p < .0001; the problem type effectwas not significant for Thai and U.S. blackyouth. United States whites had significantlyhigher ratings for undercontrolled than over-contxolled problems (means: 0.34 and 0.26);Embu youngsters showed the reverse pat-tern (0.36 for overcontrolled, 0.15 for under-controlled).

Viewing tbe interaction from anotherperspective, the effect of culture was sig-nificant for both overcontrolled problems,F(3, 300) = 21.31, p < .0001, and undercon-trolled problems, F(3, 300) = 16.13, p <.0001. On overcontrolled problems, tbeEmbu mean (0.36) was significantly higherthan the means for Thais (0.20), U.S. whites(0.26), and U.S. blacks (0.19) (all p's <.0001), and U.S. whites were bigher thanThais (p < .01). On undercontrolled prob-lems, U.S. whites (0.34) were rated higherthan Embus (0.15) and Thais (0.19) (both p's< .0001), with no other pairwise compari-

sons significant. The interaction is shown inFigure 1.

Secondary analysis excluding somaticproblems.—Particularly salient in the find-ings reported above were the high levels ofsomatic problems reported for Embu young-sters. We sought to learn how much of theculture X problem type interaction might beaccounted for by such problems. To do this,we recomputed tbe 4 x 2 x 2 (culture xgender x problem type) repeated-measuresANOVA, but with the 21 problems that in-volved somatic complaints excluded. Asmight be expected, the results changedsubstantially. Two effects were significant:(1) a gender x problem type interaction(F[l, 300] = 8.82, p < .005) closely resem-bling the interaction in the original analysis(see above), and (2) a culture x problemtype interaction (F[3, 300] = 8.52, p <.0001) quite different from that in the origi-nal analysis. With somatic problems ex-cluded, the four cultures showed a muchmore modest difference than before in over-controlled problems, F(l, 300) = 5.48, p <.005; U.S. whites (0.30) had higher scoresthan Thais (0.22) and Embus (0.21), with noother pairwise differences significant. Onundercontrolled problems, the culture effectremained strong, F(l, 300) = 16.13, p <.0001; U.S. whites (0.34) were rated bigherthan Thais (0.19) and Embus (0.15), with noother pairwise tests significant. Viewing theInteraction from the other direction, wefound that overcontrolled scores werehigher than undercontrolled scores for Em-bus (p < .0001) and Thais (p = .0005), but

106 Child Development

that there was no significant difference forU.S. blacks or whites.

DiscussionEmbu and U.S. white children had simi-

lar, relatively high total problem scores, butthe types of problems leading to those scoreswere evidently quite different for these twocultural groups. Embu youth were ratedmuch higher than the other three cultureson overcontrolled problems, and U.S. whiteswere rated significantly higher than Embuand Thai youth on undercontrolled prob-lems. Tbese cross-cultural differences war-rant closer examination.

First, it is useful to consider the pair-wise comparisons between Embu and othercultures. Thai culture, which is steeped inBuddhist tradition, is widely viewed asstrongly emphasizing self-control, emotionalrestraint, and inhibition (see Boesch, 1977;Gardiner, 1968; Sangsingkeo, 1969)—that is,as emphasizing what might be called "over-control"—and, accordingly, Tbai youth haverepeatedly been found in previous researchto have higher levels of reported overcon-trolled problems than American youth of thesame ages (see Weisz et al., 1987, 1989,1992). The level of overcontrol reported forThai youth here fell well below the levelreported for our Embu sample. (Notably, themean for overcontrolled problems was alsolower for Thais than for Caucasian-Americanyoungsters in tbe present sample. This pat-tern may have resulted from the fact that ourScimpling procedures [see Method section,above] generated a rather uniformly low-SES sample in each culture; in previouscomparisons SES has been free to vary.)

If one searches for an explanation of thehigh overcontrolled scores for Embu young-sters in the child-rearing patterns to whichthey are exposed, one finds only modest rel-evant information in tbe publisbed litera-ture. As noted in the introduction, parents ofEmbu youth rate obedience and politenessas the two things they want most in theirchildren. Those who bave written abouttraditional and more recent child rearingamong neighboring African groups in Kenya(e.g., Kenyatta, 1938; Munroe & Munroe,1977) have emphasized that children aresubjected to strong pressures toward compli-ance witb strict social rules. It is possiblethat such pressures foster the developmentof overcontrolled behavior among Embuchildren. Such pressures should also, ofcourse, discourage development of many

forms of undercontrolled behavior, and inaccord with this notion, Embu children wererated significantly lower than American chil-dren (both black and white) on undercon-trolled problems.

To examine further the large culture ef-fect on overcontrolled problems it is usefulto identify those individual overcontrolledproblems that showed the largest cross-cultural differences. Nearly all the large ef-fects (by Cohen's [1988] definition) of thistype involved problems for which Embuwere rated higher than all other groups.Some of these problems—such as no. 32"feels s/he bas to be perfect," no. 52 "feelsguilty," no. 66 "compulsions," no. 76"sleeps less than most"—sound like thekinds of difficulties one might expect of chil-dren strongly pressured to be polite, compli-ant, and well behaved. This interpretation isfurther supported by the low level of under-controlled behavior reported for Embu chil-dren relative to the two American groups,assuming that American children are givengreater behavioral latitude tban are Embuchildren.

It is especially important to note, how-ever, tbat most of tbe really large culture ef-fects involving high scores for Embu chil-dren on overcontrolled problems were forproblems involving somatic concerns, suchas no. 56a "aches or pains," no. 56b "head-aches," no. 56c "nausea, feels sick," no. 56d"problems with eyes," and no. 56g "vom-iting." When the 22 problems involving so-matic elements were excluded from theanalyses, the culture difference on overcon-trolled scores was no longer significant. Thisindicates that this particular culture effectresulted primarily from group differences inparents' reports of somatic problems. Giventhis fact, special attention must be paid tothe meaning of such reports.

Parent reports of children's somaticproblems could result from a number ofcauses. First, some psychodynamic theorieshold that somatic problems can result fromconversion of psychic distress in individualswho are pressured to comply with strictlydefined behavioral rules. Second, it is possi-ble that in some societies where behavioralrules are quite strict, somatic concerns (i.e.,problems which appear to involve "illness")evoke more positive responses from adultsthan do behavioral problems, and that so-matic concerns are tbus reinforced. Third,as suggested by Embu villagers outside oursample, somatic concerns may develop in

Weisz et al. 107

youngsters who are expected to work veryhard all the time, if the apparent ailmentsare rewarded by freeing the youngsters fromwork responsibilities. Fourth, parent reportsof somatic problems could be influenced byconditions or events that heighten their at-tentiveness to such problems. For example,the Embu people we sampled had been in-terviewed frequently about health concernsin their families during a 1-year period 5years prior to our study; it is possible thatthose interviews sensitized parents to signsof physical problems in their children 5years later. Fifth, it is possible that chil-dren's somatic concerns may refiect realphysical problems, some resulting from poornutrition or other aspects of impoverishedliving conditions. For all the somatic itemsunder category no. 56 in the checklist (seeTable 1), parents were asked to indicatewhether their cbild had "physical problemswithout known medical cause," but parentsmay not always know whether their child'sproblems have a medical cause. We mustconsider the possibility that children in de-veloping countries, more than children intlie United States, may be exposed to condi-tions that put them at risk of such problemsas no. 56c "nausea," no. 56g "vomiting," andeven no. 56a "aches or pains." On the otherhand, we know of no reason to think thatsuch conditions would be more prevalent inKenya than Thailand, and ratings for Thaiyoungsters on somatic problems were notparticularly high. Nonetheless, it will be im-portant not to overinterpret the psychologi-cal significance of the findings reported hereuntil more data can be collected regardingpossible artifactual and physical causes.

In this connection, we carried out oneadditional analysis to explore whether theEmbu children received more extreme rat-ings on the somatic items than did childrenin the three other samples. We reanalyzedeach of the 21 items involving somatic con-cerns, restricting the analysis to subjectswho had received either a 1 or a 2 rating onthe checklist. Fisher's exact test was used todetermine whether the distributions of cul-ture (Embu vs. non-Embu) and level (1 vs.2) were independent of one another. Acrossthe 21 items, we found 14 for which the pro-portion of 2's was higher for Embu thannon-Embu samples. Such a tendency towarddisproportionately high ratings for Embuchildren might suggest that Embu children'sratings were more frequently associatedwith real physical ailments than were thesomatic problem ratings for non-Embu chil-

dren. However, it could certainly be arguedthat if Embu children had developed astrong tendency toward somatic expressionof psychological stress, that strong tendency,too, could be refiected in numerous ratingsof 2 on somatic items. Thus, the analysis pro-vided new data but no clear resolution.

One other issue should be raised. Largecross-cultural differences may also resultpartly from cross-national differences in in-terpretation of the questions asked. We triedto insure faithful replication of the CBCLitems by using multiple waves of translationand back-translation, as recommended bycross-cultural experts (e.g., Brislin, 1970;Draguns, 1982; Wagatsuma, 1977), but trans-lations can always be improved. The largeculture effects found here may need to betested for robustness across further adjust-ments in translation of the instrument.

The gender effects found here are note-worthy because they fit fairly well into otherfindings ou over- and undercontrolled prob-lems in boys and girls. A gender X problemtype effect showed that boys and girls hadidentical overcontrolled scores, but thatboys showed higher undercontrolled scoresthan girls. These findings combine withthe results of numerous other studies (seeAchenbach et al., 1990; Rutter & Garmezy,1983) to suggest that boys are more likelythan girls to develop undercontrolled prob-lems, that this trend may be rather robustacross different cultures and different social-ization practices, and that it may thus resultfrom relatively culture-transcendent causalforces. Such forces might include very wide-spread gender differentiation in the social-ization of boys and girls or biological/temperamental differences. To weigh suchpossibilities we will need continued assess-ments of early temperament, and of childand adolescent behavior problems, amongboys and girls of diverse cultures.

Finally, a reminder is in order that thedata presented here were all derived fromparent reports on child behavior. Althoughparent reports may well be our best singlesource of information on child problems, itmust also be true that what parents notice,and what they consider to be a significantproblem in their child, can be infiuenced bycultural ethos and the social values of theparents' society. It will be important, in allresearch of the sort represented by thisstudy, to extend the assessment of children'sproblems to informants other than thechild's parents (e.g., teachers, the children

108 Child Development

themselves—informants who also share theculture's values) and to include direct obser-vations of child behavior where feasible. Aswe noted in tbe introduction, tbe actual andthe perceived are, to a certain extent, insepa-rably interwoven in reports of child behaviorand child problems by any informants (in-cluding trained observers). Given the over-determined nature of the phenomena—thefact that "child problems" inevitably involvea combination of actual child behavior andthe lens through which that behavior isviewed—a thorough triangulation of infor-mants and methodologies will be needed togenerate a fully informative base of cross-cultural data.

ReferencesAchenbach, T. M. (1991). Manual for the Child

Behavior Checklist/4-18 and 1991 Profile.Burlington: University of Vermont, Depart-ment of Psychiatry.

Achenbach, T. M., & Edelbrock, C. S. (1978). Theclassification of child psychopathology: A re-view and analysis of empirical efforts. Psycho-logical Bulletin, 85, 1275-1301.

Achenbach, T. M., & Edelbrock, C. S. (1981). Be-havioral problems and competencies reportedby parents of normal and disturbed childrenaged four through sixteen. Monographs of theSociety for Research in Child Development,46(1, Serial no. 188).

Achenbach, T. M., & Edelbrock, C. S. (1983).Manual for the Child Behavior Checklist andRevised Child Behavior Profile. Burlington:University of Vermont, Department of Psy-chiatry.

Achenbach, T. M., Hensley, V. R., Phares, V., &Grayson, D. (1990). Problems and competen-cies reported by parents of Australian andAmerican children./ournaZ of Child Psychol-ogy and Psychiatry, 31, 265-286.

Achenbach, T. M., Verhulst, F. C , Baron, G.D.,& Akkerhuis, G. W. (1987). Epidemiologicalcomparisons of American and Dutch chil-dren: I. Behavioral/emotional problems andcompetencies reported by parents for ages4 to 16. journal of the American Academyof Child and Adolescent Psychiatry, 26,317-325.

Boesch, E. (1977). Authority and work attitudes ofThais. In K. Wen & K. Rosenberg (Eds.), Thaiin German eyes (pp. 176—231). Bangkok:Kledthai.

Brislin, R. W. (1970). Back translation for cross-cultural research. Journal of Cross-CulturalPsychology, 1, 185-216.

Cohen, J. (1988). Statistical power analysis forthe behavioral sciences (rev. ed.). New York:Academic Press.

Dhadphale, M., & Ibrahim, B. (1984). Learningdisabilities among Nairobi school children.Acta Psychiatrica Scandinavica, 69,151—155.

Draguns, J. G. (1982). Methodology in cross-cultural psychology. In I. Al-Issa (Ed.), Cul-ture and psychopathology (pp. 33—70). Balti-more: University Park Press.

Gardiner, H. W. (1968). Expression of angeramong Thais: Some preliminary findings.Psychologia, 11, 221-228.

Hetherington, E. M., & Martin, B. (1986). Familyinteraction. In H. G. Quay fie J. S. Werry(Eds.), Psychopathological disorders of child-hood (3d ed., pp. 332-390). New York: Wiley.

Hollingshead, A. B. (1975). Four-Factor Index ofSocial Status. Unpublished manuscript, YaleUniversity.

Kenyatta, J. (1938). Facing Mount Kenya: Thetribal life of the Gikuyu. London: Martin,Seeker & Warburg.

Lambert, M. C, Weisz, J. H., & Knight, F. (1989).Over- and undercontrolled clinic referralproblems of Jamaican and American childrenand adolescents: The culture-general and theculture-specific. Journal of Consulting andClinical Psychology, 57, 467-472.

Levine, R. A., & Levine, B. B. (1977). Nyansongo:A Gusii community in Kenya. Huntington,NY: Krieger.

Mitchell, S., & Abbott, S. (1987). Gender andsymptoms of depression and anxiety amongKikuyu secondary school students in Kenya.Social Science and Medicine, 24, 303-316.

Montenegro, H. (1983). Salud mental del escolar:Estandarizacion del inventario de problemasconductuales y destrezas sociales de T. Ach-enbach en ninos de 6 a 11 anos. Santiago,Ghile: Gentro de Estudios de Desarrollo y Es-timulacion Psisocial, Roman Diaz 26, Of. 63.

Munroe, R. L., & Munroe, R. H. (1977). Goopera-tion and competition among East African andAmerican children./ourna? of Social Psychol-ogy, 101, 145-146.

Rutter, M., & Garmezy, N. (1983). Developmen-tal psychopathology. In E. M. Hetherington(Ed.), P. H. Mussen (Series Ed.), Handbookof child psychology: Vol 4. Socialization, per-sonality, and social development (pp. 775-911). New York: Wiley.

Saberwal, S. (1970). The traditional political sys-tem of the Embu of central Kenya. Kampala,Uganda: Makerere Institute of Social Re-search.

Sangsingkeo, P. (1969). Buddhism and some ef-fects on the rearing of children in Thailand.In W. Gaudill & T. Y. Lin (Eds.), Mentalhealth research in Asia and the Pacific (pp.286-295). Honolulu: East-West Genter Press.

Sigman, M., Neumann, G., Garter, E., Gattle,D. J., D'Souza, S., & Bwibo, N. (1988). Home

Weisz et al. 109

interactions and the development of Embutoddlers in Kenya. Child Development, 59,1251-1261.

Sigman, M., Neumann, G., Jansen, A. A. J., &Bwibo, N. (1989). Gognitive abilities of Ken-yan children in relation to nutrition, familycharacteristics, and education. Child Devel-opment, 60, 1463-1474.

Stanger, G., McGonaughy, S. H., & Achenbach,T. M. (in press). Three-year course of behav-ioral/emotional problems in a national sampleof 4- to 16-year-olds: II. Parent's ratings, fam-ily variables, and stressful experiences aspredictors of syndromes./owrraaZ of the Amer-ican Academy of Child and Adolescent Psy-chiatry.

Suvannathat, G. (1979). The inculcation of valuesin Thai children. International Social ScienceJournal, 31, 477-485.

Suwanlert, S. (1974). Some personality character-istics of Thai students. In W. P. Lebra (Ed.),Youth, socialization, and mental health inAsia and the Pacific (pp. 75—84). Honolulu:University Press of Hawaii.

V/agatsuma, H. (1977). Problems of language incross-cultural research. Annals of the NewYork Academy of Science, 285, 141—150.

V/eisz, J. R. (1989). Gulture and the developmentof child psychopathology: Lessons from Thai-land. In D. Gicchetti (Ed.), The emergence ofa discipline: Rochester symposium on devel-opmental psychopathology (Vol. 1, pp. 89-117). Hillsdale, NJ: Erlbaum.

Weisz, J. R., Suwanlert, S., Ghaiyasit, W., Weiss,B., Achenbach, T. M., & Eastman, K. L.(1992). Behavioral and emotional problemsamong Thai and American adolescents: Par-ent reports for ages 12—16. Manuscript sub-mitted for publication.

Weisz, J. R., Suwanlert, S., Ghaiyasit, W., Weiss,B., Achenbach, T. M., & Trevathan, D. (1989).Epidemiology of behavioral and emotionalproblems among Thai and American chil-dren: Teacher reports for ages 6-11. Journalof Child Psychology and Psychiatry, 30,471-484.

Weisz, J. R., Suwanlert, S., Ghaiyasit, W., Weiss,B., Achenbach, T. M., & Walter, B. A. (1987).Epidemiology of behavioral and emotionalprohlems among Thai and American chil-dren: Parent reports for ages &-11. Journal ofthe American Academy of Child and Adoles-cent Psychiatry, 26, 890-898.

Weisz, J. R., Suwanlert, S., Ghaiyasit, W., Weiss,B., Walter, B., & Anderson, W. W. (1988). Thaiand American perspectives on over- and un-dercontrolled child behavior problems: Ex-ploring the threshold model among parents,teachers, and psychologists. Journal of Con-sulting and Clinical Psychology, 56,601—609.

Whiting, B. B., & Whiting, J. W. M. (1975). Chil-dren of six cultures: A psycho-cultural analy-sis. Gamhridge, MA: Harvard UniversityPress.