Response Bias and Gay Politics in California, or I Can’t Believe I’m Using Something I Learned in POLISCI 2

![](http://election.princeton.edu/content/images/2008/09/wilder.jpg "Bradley Effect")

Although Princeton's Dan Hopkins shows the gap between polls and election results is narrowing, an effect clearly existed as late as the 1990's.

It seems like every morning in my news sources anymore (if you’re interested in California politics, you should definitely be skimming through [Rough & Tumble](http://www.rtumble.com)), there are a whole bunch of articles chronicling the federal trial in San Francisco that will determine [whether Proposition 8 is discrimination that violates the U.S. Constitution](http://www.mercurynews.com/bay-area-news/ci_14265947?source=rss) (most recently, [SJ Merc](http://www.mercurynews.com/bay-area-news/ci_14265947?source=rss), [LA Times](http://latimesblogs.latimes.com/lanow/2010/01/witness-cites-wide-support-for-gays-and-lesbians.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+lanowblog+%28L.A.+Now%29&utm_content=Google+Reader), [OC Register](http://totalbuzz.freedomblogging.com/2010/01/25/prop-8-trial-the-political-power-of-gays/28743/), [Sac Bee](http://www.sacbee.com/topstories/story/2489193.html#mi_rss=Top%20Stories), [SF Chron](http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2010/01/25/BAAE1BNF3G.DTL&feed=rss.news), and [WSJ](http://online.wsj.com/article/SB20001424052748704762904575025481079390028.html#mod=todays_us_page_one)). Prop. 8, you’ll remember, was the initiative on California’s ballot in 2008 that [banned gay marriage in the state](http://en.wikipedia.org/wiki/Proposition_8).

Yesterday, the defense, hoping to uphold Prop. 8 under federal law and keep gay marriage banned, called their first witness: Claremont McKenna College political scientist Kenneth Miller. Miller argued that gays and lesbians are not, in fact, discriminated against, citing their wide support from the Democratic Party and a number of unions and corporations. Rather, Prop. 8 is the exception to the rule, a deviation from a number of measures that support gays and lesbians and a number of measures that have failed without gay backing.

Miller’s argument is especially important to the defense because the plaintiffs have relied on the Prop.8 campaign as appealing to anti-gay biases to make their case. In this light, I found Miller’s testimony particularly striking, considering that just last week Stanford University political scientist Gary Segura testified that gays are in fact subject to discrimination (reported here in The Stanford Daily).

Now, I don’t want to take Segura’s side just because of our shared educational institution, but the contradiction between Segura’s and Miller’s testimonies got me thinking. Indeed, the voters of California spoke when they passed Prop. 8 by a 52%-48% margin, cutting across a number of demographics. Is it possible that the differences between whether gays have clout or not, are discriminated against or not, and between Segura’s and Miller’s different testimonies, be explained through response bias?

Response bias, which you learn about in POLISCI 2, Stanford’s basic, introductory course in American politics, occurs when “respondents answer questions in the way they think the questioner wants them to answer rather than according to their true beliefs” (thanks, Wikipedia). In political science and voting, this means that a polling respondent or someone asked in public may answer a question differently than they would in the privacy of the ballot box in order to look good.

There’s history of this happening with black candidates. Racists, who don’t want to admit they’re racists, tell pollsters and other people that they support the black candidate, only to vote against him or her later when no one can confirm how they voted. For a while, this resulted in black candidates polling significantly better than they actually did in election returns. The best example came in the 1982 race for California governor, which showed longtime black Los Angeles Mayor Tom Bradley leading George Deukmejian by wide margins, only to lose the election, which is why response bias is sometimes called the Bradley Effect.

Similar phenomena happened to other black politicians, including the nation’s first African American governor, L. Douglas Wilder of Virginia, New York City Mayor David Dinkins in his 1989 campaign against Rudy Giuliani, and Jesse Jackson in his 1988 bid for the Democratic nomination for president.

To be fair, there’s limited evidence of significant response bias in the 2008 election where the nation chose Barack Obama to be its first black president.

But could response bias have played a role in gay marriage politics in California? Miller could certainly be right that gays enjoy considerable clout in California, but that’s when you ask people directly and they don’t have the secrecy of the election booth. When push comes to shove, and people actually voted on whether or not to allow gay marriage in California, discrimination in the privacy of the ballot box could have tipped the balance to keep gay marriages illegal.

In other words, I’m not totally sure Miller’s testimony passes the smell test. Especially considering reports in the news, when it sounds like his evidence wasn’t particularly quantitative and, although an expert in California politics, he admitted to no knowledge of the history of gay discrimination.

If indeed behind-the-scenes, not obvious discrimination did occur that allowed Prop. 8 to pass, I hope U.S. District Chief Judge Vaughn Walker digs deep enough to consider that evidence and not just take it face value.

ASIDE: Yes, I wrote this in the Review’s blog. If that seems out of place, take a look at Ted Olson’s conservative defense of gay marriage. Olson, one of the attorneys for the plaintiffs in the Prop. 8 trial, is one of the foremost conservative legal minds in the country, successfully arguing Bush v. Gore before the Supreme Court in 2000 before becoming President George W. Bush’s solicitor general.