T his study compared parents' ratings of behavioral and emotional problems on the Child Behavior Checklist (Achenbach, 1991;Achenbach & Rescorla, 2001) for general population samples of children ages 6 to 16 from 31 societies (N = 55,508). Effect sizes for society ranged from .03 to .14. Effect sizes for gender were ≤ .01, with girls generally scoring higher on Internalizing problems and boys generally scoring higher on Externalizing problems. Effect sizes for age were ≤ .01 and varied across types of problems.Total Problems scores for 19 of 31 societies were within 1 SD of the overall mean of 22.5. Bisociety correlations for mean item scores averaged .74. The findings indicate that parents' reports of children's problems were similar in many ways across highly diverse societies. Nonetheless, effect sizes for society were larger than those for gender and age,indicating the need to take account of multicultural variations in parents' reports of children's problems.Children of immigrant parents constitute increasing proportions of populations served by mental health, educational, medical, and welfare systems in many societies. In addition, assessment of needs for child mental health services is a significant public health goal around the world. To meet these challenges for assessment of behavioral and emotional problems in diverse societies, there is a need for instruments that are easily administered, scored, and interpreted by a wide range of practitioners and researchers and that demonstrate multicultural robustness. Multicultural robustness is established through systematic research demonstrating that an instrument performs similarly across many societies in terms of features such as reliability, internal consistency, factor structure, scale scores, and associations of scores with age and gender (Geisinger, 1994).In the early stages of multicultural research, mental health specialists in a society often evaluate instruments developed in other societies for use in their own. If an instrument is in a foreign language, a translated version is created, and then an independent back-translation into the original language is done to verify that the translation captures the meaning of the original. Ideally, researchers then collect data using the instrument with a large general population sample. When data are available from many societies, they can be analyzed together to compare variations between societies and within societies. Establishing the multicultural robustness of an instrument is thus an incremental process using an etic approach to research, whereby the same standardized assessment instrument is used in different societies. This contrasts with an emic approach to research, whereby the meanings of the constructs under study are explored within each society.

There is a growing need for multicultural collaboration in child mental health services, training, and research. To facilitate such collaboration, this study tested the 8-syndrome structure of the Child Behavior Checklist (CBCL) in 30 societies. Parents' CBCL ratings of 58,051 6- to 18-year-olds were subjected to confirmatory factor analyses, which were conducted separately for each society. Societies represented Asia; Africa; Australia; the Caribbean; Eastern, Western, Southern, and Northern Europe; the Middle East; and North America. Fit indices strongly supported the correlated 8-syndrome structure in each of 30 societies. The results support use of the syndromes in diverse societies.

Parent-teacher cross-informant agreement, although usually modest, may provide important clinical information. Using data for 27,962 children from 21 societies, we asked the following: (a) Do parents report more problems than teachers, and does this vary by society, age, gender, or type of problem? (b) Does parent-teacher agreement vary across different problem scales or across societies? (c) How well do parents and teachers in different societies agree on problem item ratings? (d) How much do parent-teacher dyads in different societies vary in within-dyad agreement on problem items? (e) How well do parents and teachers in 21 societies agree on whether the child's problem level exceeds a deviance threshold? We used five methods to test agreement for Child Behavior Checklist (CBCL) and Teacher's Report Form (TRF) ratings. CBCL scores were higher than TRF scores on most scales, but the informant differences varied in magnitude across the societies studied. Cross-informant correlations for problem scale scores varied moderately across societies studied and were significantly higher for Externalizing than Internalizing problems. Parents and teachers tended to rate the same items as low, medium, or high, but within-dyad item agreement varied widely in every society studied. In all societies studied, both parental noncorroboration of teacher-reported deviance and teacher noncorroboration of parent-reported deviance were common. Our findings underscore the importance of obtaining information from parents and teachers when evaluating and treating children, highlight the need to use multiple methods of quantifying cross-informant agreement, and provide comprehensive baselines for patterns of parent-teacher agreement across 21 societies.

The construct representation of the cross-informant model of the Child Behavior Checklist (CBCL) and the Teacher Report Form (TRF) was evaluated using confirmatory factor analysis. Samples were collected in seven different countries. The results are based on 13,226 parent ratings and 8893 teacher ratings. The adequacy of fit for the cross-informant model was established on the basis of three approaches: conventional rules of fit, simulation, and comparison with other models. The results indicated that the cross-informant model fits these data poorly. These results were consistent across countries, informants, and both clinical and population samples. Since inadequate empirical support for the cross-informant syndromes and their differentiation was found, the construct validity of these syndrome dimensions is questioned.

No abstract