This proposal addresses an important survey data quality issue: the accuracy of self-reported divorce data i surveys and the factors that lead to under-reporting of divorce in survey data. Even though divorce is an event with high salience for individuals, measurement of divorce is problematic and, for males, so fraught with measurement error' that tahe data are often considered unusable for analyses of trends or levels of marital disruption. To address the sources of these errors, we will draw a sample of 3,600 men and women from divorce records contain high quality data on other characteristics of the divorces that will permit detailed analyses of nonresponse effects. We will compare information taken from divorce certificates against survey reports obtained by mail questionnaire, telephone interview, and personal interview. We will address four questions; What proportion of divorced men and women fail to report their divorce in a survey? What are the antecedents of failure to report a divorce? How does report reliability differ by survey mode? Does report reliability for divorce decrease or increase with time? The answers to these questions will inform future research design and protocol regarding the survey mode that provides the best communication setting for obtaining the most accurate reports of divorce and will provide a critical sense of scale for the extent of and differentials in underreporting for the analysis of survey data on marital stability. The practice of dropping men has been based on report discrepancies that are not well understood and have not been adequately researched. The research community and the development of social policies concerning families would be better served if we developed an understanding of the sources of error and ways to avoid and correct for athe errors in the future.
