In most attempts to study the prevalence of psychiatric disorder, cases have been defined by applying clinical judgment to symptom data. In some of these, such as the Midtown and Stirling County studies, this has involved assessments of written records, consisting for the most part of data collected by lay interviewers. Findings from the present study (in the Washington Heights section of New York City) indicate that respondents from the general population are less likely to be rated cases on the basis of actual interviews by psychiatrists than on the basis of interview protocols. Moreover, this difference varies with the type of interview and with the social status of the respondent.Abstract Teaser