This paper introduces a simple approach for assessing which survey questions are more likely to elicit political identity-influenced responses. We use daily data from Gallup to test which survey self-reports exhibit more or less susceptibility to politicization, finding the highest likelihood of politicization for societal-level questions. Conversely, we show that self-reported assessments of personal finances are less sensitive to partisan motivated responding. We also show how egotropic economic evaluations are influenced by the presence of other items on the same survey. Taken together, our results uncover scope conditions for how to interpret self-reported views of the economy, and we argue that measures of public opinion which have not yet been strongly politicized are better proxies for capturing the underlying welfare of the public.

Misinformation about events surrounding the 2020 election and the COVID-19 pandemic pose an existential threat to American democracy and public health. Public opinion surveys reveal that high percentages of Republicans indicate that they endorse some aspects of mistaken beliefs surrounding election fraud in the 2020 election. Still, understanding how to measure the endorsement of misperceptions is critical for understanding the threat at hand. Are high levels of mistaken beliefs genuinely held, or are they partially a function of expressive responding? I address this question through a set of survey experiments encouraging accuracy-oriented processing among the general public. Using well-powered surveys of Republicans and Independents, I find that treatments designed to encourage more accurate responses are ineffective in reducing the endorsement of partisan electoral and public health misperceptions and can in some cases even backfire. These findings suggest that support for these misperceptions is genuinely held.