With the number of women running in high-profile races on the rise in recent years, a new study raises questions about whether political polls accurately state their electoral prospects, John Sides reports at FiveThirtyEight, the New York Times blog on politics and polling.
After studying more than 200 gubernatorial and U.S. Senate races in over 40 states, political scientists Christopher Stout and Reuben Kline concluded that “pre-election polls consistently underestimate support for female candidates when compared to white male candidates.” The tendency is more common in states where more conservative views on gender issues prevail, they said, suggesting that voters there are reluctant to acknowledge support for a woman even if they intend to vote for her.
While other studies have shown that women actually fare as well as men in actual voting, Sides says, “reluctance among citizens to express their support for a woman candidate… certainly does little to encourage women to run.”
In last year’s U.S. Senate race, pre-election polls turned out to be quite accurate in predicting Republican Linda McMahon’s performance. The Quinnipiac Poll forecast her winning 44 percent of the vote, and Rasmussen gave her 43 percent; she actually received 43.24 percent of the vote.
In 2006, however, the Q-Poll forecast then-Gov. M. Jodi Rell’s winning 59 percent of the vote in her re-election effort; she actually got 63.2 percent–a 4.2 percent difference, and 1 point more than the margin of error.