Feasibility of validating survey self-reports of mental health service use

Survey respondents may inaccurately report mental health service utilization for motivational and cognitive reasons. There is little evidence on accuracy of self-reports of mental health service use, and this evidence suggests that respondents tend to underreport inpatient utilization. This study addressed the question of self-report accuracy by comparing survey data from a large probability sample to data from computerized records of publicly funded mental health services. Few inaccuracies in self-reporting were detected. However, despite the use of data bases that were unusually appropriate for verifying self-reports, several problems limited the feasibility of validation. Suggestions are offered for increasing the feasibility of validation in future studies.     

GENDER DIFFERENCES IN DEPRESSIVE SYMPTOMS

Large community studies of depressive symptomatology provide mixed evidence concerning gender differences. The present paper investigates the effect of the high skewness typically present in distributions of depressive symptom scores on findings of a gender difference in depression. Because of this skewness, a few extreme scores among women (or in any subgroup) can produce a significant between-groups difference in untransformed scores even when the two groups' distributions are otherwise similar. Data from a community survey ( n = 1316) are consistent with this hypothesis. The data also suggest that gender increases depression scores indirectly, rather than directly, for the few women scoring in the extreme upper tail of the distribution. It is concluded that the gender difference in depressive symptoms lacks robustness, and that when it does occur, the effect of gender on depression can be understood as ndirect, i.e., as mediated by gender differences in rate of employment, job status, education, and income.