Charting Obama’s Shrinking Support

When it comes to getting re-elected in 2012, President Obama is in real trouble. No matter how you divide the electorate, support for the president is down across the board, even with groups of voters who have been fanatically loyal since his election.

I meet people every day who once supported Obama but don’t anymore. I have not yet met a single person who opposed Obama in 2008 but has now come around to support him. This made me wonder: Is there anyone who is more motivated now to vote for Obama than they were in 2008? It doesn’t look like it. I could find no group of voters that has increased its support for Obama since 2008, and every group has dropped in its support.

My methodology was pretty simple: I compared 2008 exit polling data to current polling, broken out by demographic groups. There was no single source for this data but, when I could, I relied on major polling firms like Gallup, Pew, and Rasmussen. The following charts show the results, listed in descending order based on the groups that most supported Obama in 2008.

	2008	2012	Pt. Drop	% Drop
Black	95%	87%	8%	8%
Jewish	78%	65%	13%	17%
Latino	67%	58%	9%	13%
Asian	62%	43%	19%	31%
Catholic	54%	40%	14%	26%
Independent	52%	35%	17%	33%
Protestant	45%	43%	2%	4%
White	43%	37%	6%	14%

*2008 and 2012 data points are linked to their sources

Observations

• Support among Protestants has been the most consistent with only a two percentage point drop. This is also an indication that Obama was near his floor with Protestants in 2008 and didn’t have far to fall.
• Asians represent the greatest percentage point drop at 19 points.
• The largest drops in percentage terms were with Independents (33%), Asians (31%), and Catholics (26%).
• Obama’s strongest support has remained with the demographic groups that make up the smallest portion of the electorate: Blacks 12.1%, Latinos 7.4%, and Jews 4%. In other words, his strength lies with a small portion of the electorate.

Expected Criticisms

• Job approval is not the same as votes    Absolutely true, so comparing actual votes to job approval is not a direct comparison. Nonetheless, until actual votes are cast, polling data is the best proxy we have.
• You cherry-picked groups that dropped and ignored those that increased support    I included every group that I could find in both 2008 and 2012. For example, we have exit polling data for Unaffiliated Religious voters and Gays, but I couldn’t find nationwide polling that shows where these voters stand in 2012.
• Polling companies use different methodologies, so you can’t mix the data this way    I agree. In particular, I used some polling data from last week, and other polling from last March. I don’t claim that the data above will hold up to peer-review, but I think it is a valid demonstration of President Obama’s falling support across the board.
• The polling numbers do not predict the election results    Once again, this is true. Polling depends in large part on voter turnout models, and the best polls depend on “likely voters.” Most job approval polls look only at “adults” or “registered voters,” neither of which is a reliable prediction indicator.