Daniel W. Drezner
An open question about the Bradley effect
A lot of poli sci bloggers have linked to this Daniel Hopkins paper that argues that the Bradley effect disappeared in 1996. His punch line: This paper compiles polling and election data for all black and female candidates for Governor or U.S. Senator from 1989 to 2006. These 249 observations from 133 elections show that ...
This paper compiles polling and election data for all black and female candidates for Governor or U.S. Senator from 1989 to 2006. These 249 observations from 133 elections show that
there was indeed a Wilder effect, but one that was specific to a particular group and political context. African Americans running for office before 1996 performed on average 2.7 percentage points worse than their polling numbers would indicate. Yet this effect subsequently disappeared. Although precision is limited because there were only 47 observations for 18 elections with black candidates in this period, these findings accord with theories of racial politics emphasizing the information environment. As racialized rhetoric about welfare and crime receded from national prominence in the mid-1990s, so did the gap between polling and performance. Even over short periods of time, the inuence of race on electoral politics can shift markedly.
Now, I really hope Hopkins is correct, for a variety of reasons.* Here’s my question, however — does evidence of a disappearing Bradley effect in state-level elections automatically imply that it has disappeared at the presidential level?**
This is a genuine question — I really don’t know. I can see valid reasons for saying that polling effects at the senatorial and gubernatorial level would translate to the presidential level. On the other hand, this is the first time an African-American has appeared this high up on their ballot; I have to wonder if the Bradley effect is most powerful the first time an African-American runs for a particular office. On the fourth hand, the Bradley effect is not the same thing as racism (I think it’s pretty clear that the latter effect hasn’t gone away) — and I don’t have a good answer for why the Bradley Effect would disappear at lower-level elections and not at the highest tier.
In other words, I don’t know. This seems like an excellent question to throw open to comments, however.
*The most obvious thing is that the more closely final results hew to exit polls, the fewer conspiracy theories that float around after the election.
**Hopkins has data from the 2008 primary that says yes, but let’s face it, Barack Obama vs. Hillary Clinton is not the same thing as Barack Obama vs. John McCain.