Quantitative Analysis - Are Ferraro’s Claims Accurate?
Much has already been said about Geraldine Ferraro’s controversial comments implying that Obama wouldn’t be where he is if he were white (or a woman). The purpose of this post is not to attack or defend her comments, but to test her hypothesis that Clinton has been hampered by her gender while Obama has been helped by his race by looking at the exit poll numbers for people who said that a candidate’s race or gender had any influence on their vote.
This analysis is limited to voting behavior, and does not assess the impact of race or gender on other aspects of the primary process, such as media coverage, fundraising, or endorsements.
The table below compares Obama’s performance among those who said race was the most important or one of several factors in their vote to those who said it made no difference.
As you can see, Obama did 1 point worse on average* among those who said race was a factor than he did among those who said it made no difference. In some states, he did as much as 30% worse among those who said race mattered than among those who said it didn’t. In the primaries so far (excluding those for which exit polls were not conducted or did not ask about the importance of gender), Obama did better among those who said they cared about race than those who said they didn’t in 10 states, and worse in 13. People who said race was a factor made up about 19 of the primary voters on average.*
Now let’s see how Clinton did among those who said gender was a factor (about 21% of primary voters on average*):
Clinton did an average of 32 points* better among those who said gender was a factor in their decision than among those who said it wasn’t.
Being a woman hurt Clinton in Alabama and Mississippi, but she did better everywhere else among people who said gender was a factor than people who said it wasn’t. In some states, the people who said gender was important made the difference. For example, look at California. Among people who said gender had no impact on their vote, Clinton and Obama were tied, 47-47. Clinton won the people who said gender was a factor by more than 3-1. Combine that with the fact that Obama did 14 points worse among people who said race was a factor than among those who said it wasn’t, and it is likely that if nobody cared about race or gender, Obama would have won California.
Of course, this just examines the self-reported impact of race and gender in people’s votes, and probably understates the actual impact. People of all races and genders might be reluctant to admit they are voting for or against a candidate because of the candidate’s race or gender. But if people who admit that race or gender played a role in their decision are any guide, Clinton is benefitting a lot more at the ballot box from being a woman than Obama is benefitting from being black.
* Averages are computed with each state getting equal weight, regardless of population. This analysis is not weighted for turnout or population and makes no attempt to quantify the impact of race and gender on the overall popular vote.
This is a very good analysis - simple, numbers based, and pretty much irrefutable. The Clinton cult of femininity is much stronger than any racial advantage/disadvantage that Obama has. Obama’s success in Iowa, Wisconsin, Illinois, Idaho, Wyoming, Utah, Washington, Colorado, Vermont, Maine, Connecticut, Delaware, Minnesota, and North Dakota seems to suggest that Obama does fairly well with white voters too.
Maybe Geraldine Ferraro can count $3 trillion dollars.
http://acropolisreview.com/2008/03/three-trillion-dollar-iraq-war-stiglitz.html
Why did you leave out South Carolina, Maryland, and Virginia
in your analysis of states where race was a factor?
Without these three states, your analysis is incomplete
at best, and appears to be biased.
By including these three states, where race was found to
be a very significant factor according to the exit
polls that I saw, it seems that you would
have to reverse your conclusion about the
overall impact of race on Obama’s
campaign.
As it stands, this analysis looks
like a very sneaky way to get
a conclusion favorable to Obama.
There are “lies, damned lies, and statistics.”
Unfortunately, the exit polls in South Carolina, Maryland, and Virginia did not ask about the importance of race and gender. They asked instead if the country was ready to elect and African American or a woman president. This is a different question, and I felt it would be comparing apples to oranges if I tried to equate that for the table.
I will edit the post to more clearly explain why those states were omitted.
I did the math using the following numbers for voter turnout:
Alabama 536,293
Arizona 455,635
Arkansas 314,234
California 4,947,074
Connecticut 354,495
Delaware 96,374
Georgia 1,060,851
Illinois 2,038,614
Louisiana 384,346
Massachusetts 1,246,771
Mississippi 420,751
Missouri 825,090
New Jersey 1,108,322
New Mexico 148,740
New York 1,862,445
Ohio 2,233,156
Oklahoma 417,207
Rhode Island 186,439
Tennessee 623,284
Texas 2,868,454
Utah 124,396
Vermont 154,960
Wisconsin 1,113,285
TOTAL 23,521,216
Using the data from the exit polls, I calculated the difference between the actual votes and what the votes would have been if everyone had ignored race and gender. Here are the results:
The total gain due to voters who thought race was important:
Clinton gains 24,683 (0.1%)
Obama gains 5,650 (0.0%)
Other loses 30,333 (0.1%)
This suggests that Obama took votes from "other" because of race, and Clinton took votes from Obama because of race. But the overall effect is minimal. Basically, a nearly equal number of people voted against Obama as voted for him because of race.
The total gain due to voters who thought gender was important:
Clinton gains 830,308 (3.5%)
Obama loses 792,574 (3.4%)
Other loses 37,734 (0.2%)
Clinton takes a significant number of votes from everyone due to her gender.
RealClearPolitics has Obama's popular vote lead as 49.5 - 46.9. If this rough analysis holds up for the rest of the states, it would imply that if Obama and Clinton had the same gender he would be leading by 53.0 - 43.5. If they had the same race, he would be leading 53.0 - 43.4.
Of course, this is all based on exit polls so take it with a big grain if salt.
Wow, that’s great! Thanks so much for taking the time to compute the actual numbers.
No problem. I was curious. The last sentence should read “If they also had the same race, he would be leading 53.0 - 43.4″
If they had the same race, but not the same gender, he would be leading by approximately 49.5 - 46.8.
[…] Purple State Pundit tries to test her claim, and finds that more voters say they decide on gender than race lines. (Though…can you really trust what people say?) […]