Nashville News Blog

Will race play a meaningful role in the presidential campaign end game? People throw around the term "Bradley-Wilder effect" as shorthand for contests where a black candidate does noticeably better in pre-election polls than at the ballot box on election day. The "effect" is commonly mischaracterized as white people telling pollsters they support the black candidate but then voting differently. But in the few cases where a white-black candidate gap diminished significantly on election day, the data suggest it's less about people fibbing to pollsters about plans to vote black than about late deciders who disproportionately break white at the end.

Ergo a key question for Election 2008: Are late deciders more likely to break for McCain? University of Massachusetts political scientist Brian Schnaffer has a worthwhile post up today at Pollster.com tackling that question with an analysis of exit poll data from Tennessee's Ford-Corker Senate race in 2006. Schnaffer shows that Harold Ford Jr. did better among late deciders than early deciders not just among all voters, but among all whites (by eight percentage points), less educated whites, rural whites, and eastern Tennessee whites. Schnaffer cautions that a midterm Senate election is a bit of a different animal than a presidential race. But, he adds, "if you believe the comparison, then the experience from Tennessee in 2006 would suggest that there is little reason to expect late deciders to break against Obama because of his race." By doing better among late deciders, Ford ended up finishing a few points closer than some pre-election polls suggested. "If a similar dynamic works for Obama," Schnaffer concludes, "he may win by a larger, not smaller, margin than the current polling suggests."