Articles and Analysis


Demographic Groups and Votes, 2008

A first look at how groups lined up in the presidential election.  Almost all groups shifted in a Democratic direction. In that sense the Obama win as very much "across the board".  The exceptions: Small towns, late deciders and (interestingly) gays. Gay support was still highly pro-Democratic, just not has high as it was in 2004. (Exit polls found 4% self-identified as gay, lesbian or bisexual and the vote split was 70-27 for Obama.)

There were few significant shifts in the relative ordering of group support, which is some evidence that no substantial restructuring of electoral alignments. If such a restructuring is to occur, it will be during the next four years, not from Tuesday's voting patterns.
For easy legibility, here is an ordering of groups based on 2008 margin:
And here are groups sorted by amount of change in the group's vote margin from 2004 to 2008:



Fascinating charts. My guess about the slight Republican uptick in the gay demographic from 2004 to 2008 is that gays who might be at all inclined to vote Republican found themselves slightly less repulsed by John McCain and Sarah Palin than by George W. Bush and Dick Cheney. That is, the Democratic vote among gays and Lesbians was unusually high in 2004 rather than unusally low in 2008.



Great chart.

There seems to be some divergence between "small towns" (which favored Mc somewhat more heavily than Bu) and "rural areas" (which were less supportive of Mc than Bu). How is each term defined?

In partial answer, I do know that McCain failed miserably to do any outreach to the agricultural community, whereas Obama had a very active agricultural advisory group. Can the difference be expained in McCain's continued opposition to ethanol and to all farm subsidies, as opposed to Obama's "open ear" policy toward agriculture?



Re: Decrease in gay democratic vote

I would also be interested in looking at the percentage of the population that self-identified as gay. Given the stigma around homosexuality and the relative ease with which someone can lie about that designation, I would be interested to see if the self-identified population had increased at all.

I would guess that the more conflicted (and thus more likely to not self identify) are more conservative ... and thus more likely to vote Republican.

Just a thought...


Tom Holbrook:


I was just wondering how I could summarize the vast exit poll data for my elections class, which meets tomorrow morning. I believe you've solved that problem for me!



Charles :

Thanks for these Informative and Useful Charts and Graphs.

Harold Meyerson in "The American Prospect" tells us that there are two groups inside the Democrat Coalition that are pretty stable and constant in Democrat Party fervor, and unlikely to abandon the Party.

The Democratic Majority Has Emerged :


I would thank if anybody reads the article and comments here.

I know that polling is an instant flash photo, and that it is contrary to long term projections, and speaking of Coalition Loyalty is very premature.

2012 und 2016 verbotten.

But anyway, Meyer's article is very intriguing. And playing long term futurology is such a fun !

And John Judis and Ruy Teixeira with the Brookings Institution saw November 4/2008, many months ago.

Vicente Duque



Great graphs. Labels a bit too light to read easily. I may use an LCD projector to show these to a class, and I'm not sure some of it will show up--a minor point considering how wonderful the data is. Thank you.



Something I'd be interested in knowing more about is voter loyalty. Once an individual has voted for a particular candidate, are they more likely to vote for them again? How likely is it that young voters who turned out to vote for Obama are going to continue voting Democrat in future elections? What about Hispanics?


Answer to todji :

I think that the answer is yes. The past votes tie you to vote again.

George Bush and Karl Rove knew that Hispanics were excellent material for the Republican Party ( conservative and religious ) and conquered a big slice of that Demography in 2004.

But Bush was harassed and gored by many problems and moderate Republicans abandoned Hispanics to the anger of the Right Wing.

Harold Meyerson predicts that they are going to stay loyal to Dems. Other forecasters are predicting that Texas, Arizona and Georgia can be moved to Dems in Future Elections ( because of Hispanics ).

Meyerson is an accurate prophet. He predicted the Obama Win in Virginia and North Carolina and other events with great anticipation.

Arnold Scwarzenegger seems "simpatico" to Hispanics and Sarah Palin gave an interview where she spoke as very moderate and Hispanic Friendly. Good for her that this was unnoticed by the right wing base.

So there are Republicans that see that there is hard work ahead.

Vicente Duque


Gary Kilbride:

Great charts, Charles. I've been cataloging the same type of thing the past few days.

The amazing relationship to me is the convergence of the rural vote and white female vote. As recently as 2000, that percentage was more than 20 points apart. This cycle it was virtually identical.


Voltron Defender of the Universe:

Great charts. The one group that might have "re-aligned" although I think that term is overused would be those earning over 200,000 a year. The one group to see an increase in their income tax after Obama's tax cut. But I wonder as a single guy earning around 25,000 clams a year will my payroll taxes go up becasue of the Obama health care plan?



What follows is pure speculation and should be taken as such.

It may be that the slight decline in support for the Democrats among gays stems from the long-term trend toward acceptance by and integration into mainstream American culture. (The Proposition 8 result in California is, I think, an aberration, much like the imposition of Prohibition in 1920.)

If that hypohesis is correct, gays may feel less constrained to vote in accordance with their status and more in accordance with their class identification.


Another Mike:

Fascinating graphs!

I think it would be incredibly interesting to look at the projected population changes in these demographic groups over the next 4, 10, and 20 years. For example, Latinos and Asians are projected to increase in voter share. Both are big Democratic groups. What about other groups, however, such as where people are projected to live (big city/small city/suburbs/small town/rural), religious affiliation, union/nonunion, etc.? Also, is it valid to assume new members of such demographic groups will vote like existing members? That seems to be the case for race, but will it hold for religion? Lots of interesting work could be done here in projecting which party is better situated to get the vote in the future based on each party's appeal to the different demographic groups of today. Just eyeballing the groups, it sure looks like the Democrats have the brighter future, but I would love to see a comprehensive analysis.



Vicente Duque and Another Mike:

My guess would be that you are right. What comes to mind is reading customer reviews online. When people have bought a new item, as long as it works to minimum specifications, they have a tendency to overrate it. So almost every review of every item ends up being 5 star best ever. In the political realm, I'd say this would mean that an independent who otherwise votes evenly between the two parties would be more likely to reelect someone.

Mike's post is where my thoughts were leading. If there is a tendency for a voter to align with a party in their first election and maintains that alignment overtime, then Republican party is in serious trouble for the near future. 68% of the youth vote went to Obama. The current generation- the baby boom echo- is the largest in US history. The latino vote went 67% to Obama. Latino's are the fastest growing segment of the population.

fivethirtyeight.com has a great piece yesterday which talked about the electoral advantage Obama started with. The conclusion was that if Obama's vote totals in EVERY state were reduced by 8.6 points across the board, he still would have won the electoral college. A 9.4 point reduction would have resulted in an electoral tie. If there is such voter loyalty and we can expect both latinos and young voters to continue voting democrat, then I'd say that this electoral advantage is locked in for at least the next few election cycles.

I'd be curious to know of any specific research and hard numbers if available.



If gays were 4% of the vote and 70% of them voted for Obama, that would translate into about 5.5 million LGBT votes for him.

Any decline in % could be due to several factors. Conservative southerner gays may be more willing to id themselves to pollsters. And then there is Obama's statement that he opposes gay marriage.

While LGBT's are furious at the GOP over its support of anti-gay referendums and obstruction of gay civil rights and hate crime law on the federal and state levels, they aren't happy that Obama stated opposition to same sex marriage.

Still, a 70 to 27% split of a group that comprised 4% of the voters is nothing to sneeze at. However, if Obama wants to retain that majority and increase it, he's going to have to deliver something to show he's on their side especially after Prop 8.



What's the sample size on the gay Republicans in the exit polling? 27% of 4% is a pretty small number. With roughly 60 subgroups listed, there should be several >2-sigma fluctuations, and the gays being more Republican this election kind of smells like one to me.


I guess Biden's statement that it was patriotic to pay taxes wasn't that controversial after all... Especially the "rich" were more than happy to vote for Obama and his increase of taxes on the $200k+: the same $200k+ moved the most to the Democratic camp (+34%).



So much has been made of Obama's supposed underperformance among gays and lesbians, when in fact, it is quite likely it is just statistical noise.

According to the exit polls, only 4% of surveyed voters are gay or lesbians. Applying the 4% to the total number of respondents, 17,836, gives a gay sample size of 713. This may even be a substantial overstatement of the gay sample size, since respondents were only asked to identify their sexuality in certain states. Generously assuming that 50% of the respondents were asked about their sexuality, this produces a gay sample size of only 357.

With a sample size of 357, the margin of error is 5%. With such a large margin of error, the change in the Democratic share of the gay & lesbian vote from 77% to 70% is just not statistically significant.

Furthermore, with so many demographic subgroups there's a good chance that the results from some groups will be off by pure chance alone.



To 4xfloor: Good post.


I've been experimenting a bit this evening. Reading about the Obama team's relationship with the "net roots," I decided to plot broadband penetration against Obama vote % and Obama % - McCain %. I did this with state data. Some correlation seems likely and the outliers make sense. See my Word face-Off blog. I'd be interested to hear some constructive criticism ;-)



The Gay uptick for the GOP was because Obama toured the south with a minister formerly "gay" who stated that all gays could be converted.
Between that and Obama's no stand on california's proposition 8 there is no place for gays in Obama's agenda.



Given sample sizes, the shift in the gay vote is well within the range of sampling error. The emphasis placed on it in this discussion cannot be supported by the data.


Vicente Duque:

Which voters are going to potentiate the Republican Party in the next Presidential Election ?

SuperConservative and Uber Republican National Review : Mark Krikorian considers latino vote as "very small mess of pottage"

But Mitch McConnell, leader of the Senate Republicans : The Republican Party can recoup Youth and Hispanics.


Vicente Duque


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.