Hurdles to Inference: The Demographic Correlates of Survey Breakoff and Shirking

Objective: Various shirking behaviors in survey-taking can inhibit researchers' ability to draw unbiased inferences from the resulting data. We define three types of shirking behavior, identify their demographic correlates among survey respondents, show how shirking may lead to biased or incorrect inferences, and offer suggestions for navigating these threats. Methods: We analyze responses from a large (N=3,256), representative survey of Americans, identify indicators of three types of shirking, and regress these indicators on survey firm administrative data on respondent characteristics to discover the respondent characteristics associated with different shirking types. Results: The data suggest women are more likely to break off during a survey, while men are more likely to shirk while completing the instrument. In addition, young, less educated respondents are more likely to use straight-lining or other satisficing tactics. Conclusion: Systematic differences in shirking behaviors across these demographic groups make us likely to misunderstand the (reported) social behavior of women and men, young and old, or more and less educated, indifferent ways. This is particularly problematic for testing behavioral theories that generate gender, age, or education-specific predictions, or, predictions regarding attitudes or behaviors that strongly correlated to these characteristics.