This project combines quantitative methods with studies using ethnographic and cognitive interviewing techniques to reconceptualize constructs and to develop and test measures in key domains affecting children in separated families: child support and other financial and in-kind transfers, contact between parents and children who live apart from each other, and information about paternity establishment for children born to unmarried parents. Analyses of existing data identify sources of error in self- reports. These analyses guide the development of improved measurement of these central variables using semi-structured interviews with, individuals and groups of respondents to refine their conceptualization and operationalization. We test revised measures in a pilot study that includes comparisons with external records from the courts. New measures are required to obtain measurement that is sensitive to changes in U.S. family structure and new social welfare policies. Techniques used here hold promise for improving the measurement of constructs with similar social, cultural, and cognitive sources of error. By developing methods for computer assisted interviewing methods, we attempt to help establish standard solutions to common measurement problems. In addition, we examine the extent to which parents report accurately about a set of their own and each others' social and economic characteristics. Analyses of existing data use two complementary sources: (1) a sample of divorce cases selected from court records in Wisconsin; (2) telephone interviews with parents in these court cases. In addition, we compare results from the analysis of proxy reports in the Wisconsin studies to those from the National Survey of Families and Households (NSFH) to assess the generalizability of these findings. The project builds on previous experience in developing and analyzing measures of child-support and paternity variables to develop survey measures that reduce systematic and random errors and improve construct validity.
