THE IMPLICATIONS OF SURVEY MODES AND METHODS IN MEASURING SOCIAL-GROUP CLOSENESS.

Social scientists have long recognized the importance of social groups in political life, exploring when and how they influence individuals' political attitudes and behaviors. But measuring group identities and psychological affiliations is problematic. Since 1972, the "group closeness" measure developed in the American National Election Studies (ANES) has been the most widely used indicator of group identification and affiliation, included in other surveys and incorporated into measures of group consciousness and community. Building on research on response effects, I develop and test hypotheses about how variations in survey methods-interview mode, presentation format of survey questions, response options, and question order-influence responses to group-closeness measures across a range of social groups. The analyses show that measurement differences generate significant response effects: Telephone interview mode and sequential format produce higher levels of reported closeness; response-option variations impair cross-survey and over-time comparisons; and question-order variation in telephone interviews generates effects consistent with satisficing, consistency bias, and social desirability. Replicating the findings from a recent publication about group closeness by re-analyzing data separately by mode and format demonstrates mode and format effects. Despite these challenges, the group-closeness measure has several measurement and substantive virtues, and should continue to be used in public-opinion surveys. The paper ends with a discussion of how best to move forward in research on group closeness: I recommend making sequential-format and scaled-response options universal to mitigate response effects in extant data and improve future measurement precision, and calibrating question order to minimize social desirability. [ABSTRACT FROM AUTHOR]