The impact of race and gender revisited
by kjblair2, Mon Jun 02, 2008 at 04:27:19 PM EDT
This is a follow up to a previous diary where I had started to look at the impact of race and gender on selected primaries after West Virginia. The premise is that one can take the data from exit polls and estimate the percentage of voters that used race and gender in their voting decisions. Most exit polls ask if race or gender was important in deciding who you voted for. By looking at how people who answered "yes" to these questions, one can determine their impact on the outcome.
More in the extended entry.
To do this, we have to make an assumption. Someone who said that gender was important and voted for Clinton would be considered to have had a positive gender influence on their vote. Likewise, if that person voted for Obama they would be considered to have had a negative gender influence. One can make the same assumption for race. That is, if race was important and they voted for Obama, that was a positive racial influence. If they voted for Clinton, that was a negative influence. This isn't to say that voters that exhibited negative gender or racial influences were necessarily sexist or racist, just that it impacted their vote.
Unfortunately, no exit polls were conducted for almost all of the caucus states. (Iowa and Nevada were exceptions.) Some of the exit polls didn't ask questions on gender and race (Nevada, Florida, Maryland and Virginia.) Finally, I didn't include Iowa, New Hampshire or South Carolina since Edwards was still a viable candidate at that point and introducing a third person into the picture made the calculations more difficult. That left me with 29 states with valid exit poll data.
What did the data show?
First, there was a lot of variability in the data. Positive gender influences ranged from 7.6% to 20.3% (i.e., people voted for Clinton because of her gender). Negative gender influences ranged from -3.0% to -19.3% (i.e., the opposite, people didn't vote for Clinton because of her gender). Positive race influences ranged from 2.6% to 19.2%. Negative race influences ranged from -4.7% to -18.0%.
Second, one can calculate a "net" influence by looking at the difference between the two numbers for each state. A positive "net" number would indicate that gender or race helped the candidate whereas a negative "net" number would show the opposite. Net gender numbers ranged from -11.2% to +15.5%. Net race numbers ranged from -15.4% to +10.4%.
Third, one can weight the results based on the number of pledged delegates allocated to each state to obtain an overall number that summarizes the impact of gender and race on the primary. (I used pledged delegates for the weighting since that had a direct impact on who won the race.) 13.1% of the weighted voters voted "for" Clinton because of her gender and 7.3% voted "against" her. 8.7% of the weighted voters voted "for" Obama because of his race and 9.3% voted "against" him. Clinton's net impact was +5.8% and Obama's net was -0.5%.
Based on the above, one could draw two conclusions. Unfortunately, there's a significant fraction of the population (slightly less than 10% but it varies from state to state) that appear to have a problem with either a woman or African-American running for President. Second, it appears that Clinton benefited more (or was harmed less) from her gender than Obama was from his race.