Who lies on surveys, and what can we do about it?

Focusing on demographic characteristics, the author seeks partially to replicate, with a larger set of variables, Belli, Traugott, and Beckmann's (2001) recent study on vote overreports using the same data from the U.S. National Election Studies (1948-1998). His analyses show that Blacks and residents of the Southern States in general are most likely to make false statements on how they voted. He suggests a possible solution for inaccuracies in survey data and proposes that, when validation of verbal responses is not possible, it may be prudent, if feasible, to re-estimate models with and without Black and Southern respondents to make sure that findings are robust.