Validity in Mental Health Surveys

Publisher Website
Google Scholar
Abstract
In recent years a method of conducting mental health surveys has been developed which relies mainly on interviewing individuals according to a questionnaire. Subjects for the surveys are chosen by statistical techniques so that they will be representative of designated populations regardless of whether or not they have ever had psychiatric treatment. The questionnaires are administered by trained interviewers who as a rule are not psychiatrists; the data are later evaluated by psychiatrists who make a series of ratings on each individual in the survey. Inasmuch as surveys of this type show a very high prevalence of psychiatric disorder in many populations (from 10% to 60%), the question of validity becomes a matter of major importance. In the present study 123 individuals were examined and rated independently by both the survey technique and by a psychiatrist employing his usual clinical approach. After the psychiatrist had made his independent rating he was allowed to see the survey data and also to re-examine the subject again if he wished. He then made a final rating constituting his best judgment based on all information available. The degree of agreement is indicated in Tables I to IV. Table I deals with symptom patterns considered without regard to diagnostic implications. Thus, a given individual may have more than one of these symptom patterns. Table II concerns impairment by reason of psychiatric symptoms. Table III deals with a rating as to whether or not the subject is a psychiatric case: ‘A’ means almost certainly a psychiatric disorder of some sort; ‘B’ means probably; ‘C’ means doubtful; and ‘D’ means almost certainly not a psychiatric case. (Note that the group here studied was deliberately weighted with ‘A’ and ‘D’ ratings and hence is not representative of any population.) Table IV summarizes the agreements. While more work is needed, and studies of this kind should be repeated, the results so far show considerable agreement between the survey and clinical methods. This suggests that the large percentage figures obtained by surveys are properly matters of serious concern and that they have far-reaching implications for the teaching and practice of psychiatry.

This publication has 2 references indexed in Scilit: