Likewise what % of respondents troll the pollsters with fake answers, and is it symmetric by party. When I see crazy results in youth crosstabs I tend to think thats whats occurring, even though basically noone in press ever mentions this possibility.
This is one of the most classic examples of social desiriability bias in surveys. Even CPS, one of the highest quality surveys in the US, finds this kind of over-reporting.
Maybe this is something people already looked at, but it just crossed my mind: who are these respondents that over-report? Are they more or less educated? Any specific racial or age group? That would be interesting to know!
Oh yeah, for sure. Hopefully it's a smaller problem than over-reporting. ANES is doing a post record linkage verification with the respondents. Hopefully that will further reduce this type of issue.
Have we decided whether the problems with *administrative* records of voting as a "ground-truth" is a substantial counterpoint? I haven't kept up on the evidence & debates. https://www.jstor.org/stable/44014601
Comments