When a null pocketbook effect is the artifact: Strategically underreported income and the attenuation of egotropic voting

We argue that the lack of a strong and consistent correlation between egotropic assessments and vote choice in observational data from poor countries may be a statistical artifact resulting from respondents' attempts to evade taxation. Using data from two rounds of the Afrobarometer, we show that respondents report significantly worse economic conditions when they think they are reporting to the government; the likelihood of misreporting increases with actual or perceived effective tax rate. Misreporting causes a conflation in the data between those who are actually experiencing worsening conditions and those who are merely claiming to, attenuating the apparent effect of pocketbook concerns on the vote. When we control for this bias in the African data, we find that egotropic evaluations are significantly and positively correlated with vote choice.