A social class dependent factor in questionnaire research

Abstract

Inordinately high numbers of respondents of low socioeconomic status (SES) agreed to certain types of questions of attitude and opinion presented to the 1660 subjects of the Midtown Manhattan Mental Health Study. SES significantly determines responses to impersonal, generalizable questions, particularly when expressed in the form of proverbs, assertive authoritarian statements, or complaints about general social conditions. Agreement distributions consistently biased by SES do not occur when the questions deal with facts related to the respondent or with opinions and attitudes linking the respondent to closely defined persons. The findings fit recent observations that language is an encoding of developmental social class experience, which determines the mode of an individual's management of social problems. The findings also emphasize the need to devise research questionnaires in language which is percieved uniformly by the diverse social classes and which evokes responses equally valid for each social class.