A major challenge facing mental health services for young people is how best to integrate important data from different informants, such as parents, teachers, and the young people themselves. To meet this challenge, the proposed project will develop and test statistical models for integrating multiple sources of data in the assessment of psychopathology. The long-range goal is to improve services by helping mental health workers make more accurate clinical decisions. Practitioners often collect information from multiple sources prior to making key decisions about the best ways to help their clients, e.g., no treatment, further evaluation, limited intervention, or one or more full-scale interventions. Many studies show disagreement among different sources of data about young people's problems. Carefully designed research is needed to optimize use and integration of data from multiple sources. Standardized instruments are important contributors to assessment, prevention, treatment, and evaluation of outcomes. Using empirically based standardized assessment instruments and DSM diagnostic interviews; the investigators will analyze data on psychopathology collected from multiple sources in national samples from the U.S., Holland, and Australia. Classification/regression tree analysis will be used to establish cutpoints on combined multi-informant data for maximizing DSM diagnostic accuracy as an external criterion. Other statistical models will include those based on variable-centered approaches (e.g., models developed to analyze multitrait-multimethod matrices for continuous observed and latent variables), as well as those based on individual-centered approaches (e.g., latent class models for categorical observed and latent variables). External validation analyses of DSM diagnoses and other important criteria will be used to test the accuracy of results obtained from internal validation analyses. The power of different statistical models to detect differences related to clinical status, DSM diagnoses, and other indices of psychopathology will be tested. The findings are expected to advance knowledge and methods for assessing youth psychopathology that will substantially improve use of essential multisource data for mental health services.
