One of my previous posts had some exit poll data that didn't seem to add up (it has far too many black voters in the sample). Maybe it was preliminary and biased towards southern states. I'm updating some of the key results in this post, mostly from here, which has exit poll data limited to "ten key states". The "key states" in the exit poll sample are: Arizona, Florida, Georgia, Michigan, Nevada, North Carolina, Ohio, Pennsylvania, Texas, and Wisconsin. The questions on religion were not broken down by state.
Smaller subsamples from these exit poll data, however, such as the much less than 1% who identified as non-white Jewish, should be taken with a grain of salt as they are subject to greater sampling error. The total sample size is 22,914 respondents.
One of the core conclusions of the previous post, that white Evangelical Christians (who make up 14% of the general population but 22% of voters) were critical to Trump's win, still holds true, however. Without them, this election wouldn't have been close. It could also be that white Evangelical Christians are a larger percentage of people in the "key ten states" in the sample, than they are in the general U.S. population.
The number in parentheses is the percentage of the total exit poll sample, and the numbers that follow are the percentage of that group voting for Trump and Harris respectively.
- White Evangelical Christians (22%) 82-17.
- White Catholics (15%) 61-35
- White Mainline Protestants (8%) 51-45
- Non-White Catholics (7%) 49-49
- White Religious "Something Else" (5%) 43-52
- Non-White Protestants (12%) 40-60
- Non-White Religious "Something Else" (5%) 32-66
- White Nones (17%) 27-71
- Non-White Nones (7%) 25-72
- White Jews (2%) 20-80
Making Sense Of The Exit Poll Sample
About 69% of voters in the exit poll sample were white. The U.S. as a whole is about 58% non-Hispanic white, but historically, non-Hispanic whites have been overrepresented among voters, so this isn't far out of line with expectations (particularly because the "ten key states" in the exit poll sample probably don't include majority-minority jurisdictions like the District of Columba, California, and Hawaii).
Just 52% were Christian. The 52% of voters who were Christian is far out of line with other estimates of the percentage of the adult U.S. population that identifies as Christian, which is about 63%. The only age group of American adults who are 52% Christian or less are 25-29 year olds. Gallup, as of 2024, estimated that 68% of Americans are Christian.
People who identified as religious but "something else" accounted for 7% of whites and 16% of non-whites. Pew and the General Social Survey (as of 2022) and Gallup (as of 2024) estimated that the number of Americans who are religious but not Christian to be 6-7% including Jews. This exit poll has a percentage twice that high (12% of voters including "something else" and "Jews"). Muslims and Hindus are also probably underrepresented among voters compared to their share of the general population, because a larger share of them than the general population are immigrants who are not yet U.S. citizens. So among voters, non-Jewish "something else" ought to be closer to 2-4%.
I suspect that most of the 10% who are "something else" (perhaps 6-8% of the total), especially among the half of them who are white, are mostly "generic" Christians not actively involved with a particular church or denomination, or are part of a denomination that is not a good fit to the Protestant-Catholic divide (e.g. Jehovah's Witnesses, Messianic Christians, Orthodox Christians, and Mormons). This interpretation fits the similarity in voting patterns between white mainline Christians (especially those who attend church less often who would support Trump less than mainline Christians overall), and religious whites who identify as "something else."
Non-white Jews were less than 0.5% of the total exit poll sample, so it isn't possible to estimate their vote in the exit poll survey with any precision, although non-white Jews were more pro-Trump than white Jews, as discussed below.
Analysis
Non-white Protestants Hispanics (mostly Pentecostal) are significantly more pro-Trump than black Protestants, and may even be more than 50% pro-Trump. Generally speaking, Protestant Hispanics in the U.S. are more conservative than Catholic Hispanics in the U.S. This may be pulling down Harris support among non-white Protestants (far more than 60% of black Protestants supported Harris, probably more like 80%+).
But pre-election polling found, in line data from previously elections, that Catholic Hispanics favored Harris over Trump 65-34, compared to a roughly 49-49 split inferred from exit polling, while white Catholics supported Harris and Trump respectively in almost exactly the percentages shown in the exit poll (i.e. 61-38). This discrepancy is puzzling, particularly because the subsample is still fairly large and the discrepancy is too large to be plausibly due to anti-Trump bias in pre-election polling generally (which was roughly three percentage points in this election).
Jews in the exit poll were predominantly white, but the small percentage who are non-white were significantly more pro-Trump than Jews who identified as white. I suspect that these are Jews who consider being Jewish/Middle Eastern to not be white and who feel that Trump would be more pro-Israel than Harris. Pre-election polling predicted much stronger Jewish support for Trump with a predicted 65% for Harris and 34% for Trump. The small sample size, and a sample in the ten "key states" that is unrepresentative of the nation as a whole, may be a factor in this discrepancy.
Pre-election polling showed more support for Trump among Evangelicals and white Catholics who attended church more often (relative to those who attended church infrequently), and less support for Trump among white mainline Christians and black Protestants who attend church more often (relative to those who attended church infrequently). Regardless of the absolute numbers supporting Trump in the actual election, this is probably true of the actual results as well.
The results by race were more or less what you would expect, with stronger support from Trump among black and Hispanic voters than in past elections, apart from an anomalously high level of support for Trump from Native American voters. This may be due to a small sample size and a sample made up mostly of whites with trace Native American ancestry and little connection to Native American culture who nonetheless chose to identify that way in the survey (something that is fairly common, especially in parts of the South that may be among the "key ten states" in the sample).
There is a slight discrepancy between the percentage of the sample that is white in the race specific figures (71%) and in the data broken out by religion (69%) which is probably simply due to rounding errors.
As expected, all non-white categories (except Native Americans and "other") favored Harris, although the Harris lead over Trump was modest among Hispanics, and whites favored Trump.
Usually, in surveys, "other" is mostly made up of mestizo Hispanics, although it might also include Middle Easterners or Jews who doesn't identify as "white." It also may include people of mixed race.
Notably, about 31% of whites in the exit poll sample identify as Evangelical Christians.
CNN breaks out the religion data by race beyond whites (but also doesn't have state by state breakdowns on religion):
Black voters:
Hispanic voters:
White voters:
White college graduates:
Whites who are not college graduates:
All Men:
All Women:
All voters:
A few gratuitous extra statistics from the overall exit polls.
Unsurprisingly, urban voters favored Harris and rural voters favored Trump. The suburbs, as always, were in between and narrowly favored Trump:
It is stunning that even 13% of LGBT voters backed Trump. About 8% of voters identified that way:
Single ladies were for Harris, single men and married people voted for Trump:
Women preferred Harris, men preferred Trump, the eight percentage point gap was smaller than many people expected it to be:
The dividing line between Trump supporters (who are older) and Harris supporters (who are younger) is in the low 40s, but senior citizens were actually evenly split:
Married people like Trump more, single people like Harris more:
Union status was a pretty minor factor.
Generally, more educated voters favored Harris, while less educated voters favored Trump. But associates degree holders were more pro-Trump than people with "some college" (I suspect that this may be a gender effect, with more associates degree holders being men and more some college voters being women):
No comments:
Post a Comment