Strategic Survey Inflation

In this paper I develop a formal model of survey response, based on question types that admit a range of numerical evaluations and allow for numerical inflation. The model predicts that survey administrators should frame the influence or importance of their survey according to assumed characteristics of the respondent population. This analysis also suggests that by limiting evaluative criteria, survey administrators might, paradoxically, measure more precise and generalizable attitudes and dispositions. I derive testable hypotheses about respondent behavior from the model and evaluate them using data from the 2008 ANES pre-election survey. The results are mixed: while the primary hypothesis – that respondents act strategically when they perceive as survey as influential – is not confirmed, I still reveal systematic
characteristics of inflation-prone respondents. I also suggest how future empirical work, including experimental research designs, might better evaluate the model developed here. Finally, I suggest how a formal model of survey response might better explain other kinds of response bias.

Note: Abstract extracted from PDF text