When Democratic presidential candidate Hillary Clinton won the 2008 New Hampshire primary despite many polls predicting an easy victory for competitor Barack Obama, explanations included the Bradley Effect, fluctuations in voter turnout, and a wave of undecided voters turning out for Clinton on Election Day. One explanation that received little attention was the possibility that Clinton’s gender may have played a role in unintentionally disrupting polling outcomes.

A 2011 study published in Political Behavior, “I’m Not Voting for Her: Polling Discrepancies and Female Candidates,” investigates why pre-election polls in contests involving female candidates often do not accurately predict actual election outcomes. One possible reason for this discrepancy is that voters may try to distance themselves from a “feminist” ideological position, especially in socially conservative areas.

Researchers compared pre-election polling data and election results for 123 races featuring at least one female candidate and 92 races featuring two white male candidates; the races took place across 215 U.S. Senate and gubernatorial races in more than 40 states between 1989 and 2008. Additional study variables included the political leanings of the state, voter turnout and the overall status of women in a given state (measured by the strength of gender-based initiatives, the percentage of working women, and the percentage of women holding elected positions).

Key study findings include:

• Two-thirds, or 66%, of pre-election polls under-predicted how well a female candidate would ultimately fare at the polls, compared to under-predicting only 43.3% of a male candidate’s performance.
• “When compared to [similar] white male candidates … female candidates performed significantly better in the final results than would be predicted by pre-election polls to the tune of about three and a half percentage points.”
• The age and gender of voters and the social context of the state accounted for approximately 25% of the gap between pre-election polling predictions and election outcomes.
• The more women working in a given state, the lower the chances are that pre-election polls will underestimate the support for a female candidate. In fact, pre-election polls for female candidates in states with a robust female labor force tended to overestimate public support by a small percentage.

The researchers conclude that “voters — and especially those living in states with more traditional views — may want to appear less supportive of female candidates…. Pollsters and researchers should be concerned about the gender of the candidate as much or possibly even more than they are about race.”

Tags: polling, Iowa/New Hampshire, women and work

By Margaret Weigel | November 3, 2011