Persnickety Validity: Comparing Construct Validity Across Civic Duty Indicators.

The central topic of this paper is the measurement of one variable, civic duty. This variable has become a mainstay in empirical research on individual voting behavior since the 1960s and it is one of the chief explanatory saviors of the rational choice model of voting. Nevertheless, the measurement of civic duty on the American National Elections Study is prima facie substandard. Question wording is one concern, as a number of scholars have demonstrated that seemingly small changes in question wording can have large effects on the answers individuals provide and the conclusions scholars draw. My primary concern is measurement validity, commonly defined as "whether a variable measures what it is supposed to measure" (King, Keohane, and Verba 1994). In survey research, the question is the instrument through which we measure a concept; if the wording is unclear, biased, or just plain bad, then any research that uses that indicator will also be flawed.In the paper, I will explain the controversies surrounding question wording and measurement validity; I will examine the existing body of work on civic duty; and I will empirical evaluate the validity of survey instruments that use different measures of civic duty in the United States and the United Kingdom. Possible explanations for the results, recommendations for further analysis, and the implications of these findings for research on voting behavior will also be discussed. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]