Separately for ages 11/2 to 5 and 6 to 18, used items for rating behavioral and emotional problems to construct (a). "top-down" DSM-oriented scales from experts' ratings of the items' consistency with Diagnostic and Statistical Manual of Mental Disorders (4th ed. [DSM-IV], American Psychiatric Association, 1994) categories, and (b). "bottom-up" empirically based syndromes from factor analyses of parent, caregiver, teacher, and self ratings (N = 14853). Both kinds of scales are scored from the same assessment instruments and are displayed on profiles normed on the same national samples. Psychometric properties were similar for both kinds of scales. Associations between counterpart scales were medium to strong. Quantified, normed DSM-oriented and empirically based scales scored from the same instruments can facilitate assessment of individuals, statistical analyses for research purposes, and integration of top-down and bottom-up approaches to deriving constructs for psychopathology.
Assessment of adult psychopathology relies heavily on self-reports. To determine how well self-reports agree with reports by "informants" who know the person being assessed, the authors examined 51,000 articles published over 10 years in 52 peer-reviewed journals for correlations between self-reports and "informants" reports. Qualifying correlations were found in 108 (0.2%) of the articles. When self-reports and informant reports were obtained with parallel instruments, mean cross-informant correlations were .681 for substance use, .428 for internalizing, and .438 for externalizing problems. When based on different instruments, the mean cross-informant correlation was .304. The moderate sizes of the correlations argue for systematically obtaining multi-informant data. National survey findings were used to illustrate practical ways to obtain and use such data.
11 January 2011 in Journal of Clinical Child & Adolescent Psychology
The special section articles demonstrate the importance of informant discrepancies. They also illustrate challenges posed by discrepancies, plus opportunities for advancing research and practice. This commentary addresses these cross-cutting issues: (a) Discrepancies affect many kinds of assessment besides ratings of children's problems. (b) Symptom models complicate challenges posed by discrepancies. (c) Informant discrepancies validly reflect different genetically and environmentally influenced aspects of children's functioning. (d) Multisource assessment is needed to test and refine diagnostic constructs. (e) Methodological issues arise from the kinds of variables that are rated and ways of evaluating agreement. (f) Assessment requires age-, gender-, informant-, and society-based norms. (g) Developmental levels affect what can be assessed, by whom, in what contexts, and for what purposes. (h) Research-based algorithms are needed for using multi-informant data to assess individuals.