Let me be more specific: If turnout and support is as high among minorities as polls indicate, and if President Obama gets at least as much support from white voters as Democratic House candidates did in the horrid 2010 midterm elections, then Romney cannot win the popular vote.
But you can play with the numbers yourself to predict Obama's margin, with your own assumptions.
It just takes three simple steps!
1. Pick the level of minority support you think Obama will get
Scenario 1 - 2008 levels (this is about what general polls show)
Scenario 2 - better than 2008 (this is what minority-specific polls show, see below)
Scenario 3 - 2004 levels (what Republicans are hoping for, or at least something close)
(Update: for example, in 2008, exit polls showed Obama winning the Hispanic vote by a 36 point margin, and in 2004 they showed Kerry winning the Hispanic vote by a 9 point margin.)
2. Pick your racial demographics
In 2008, the electorate was 74% white, according to exit polls. Since 1992, the electorate has been ~3.2 points less white each cycle, which would put us at 71% white this year. But Gallup says that won't happen this year, and 2012 will be the same as 2008. Meanwhile, Conventional Wisdom says minorities are bummed and won't turn out this year. Minority voters themselves say otherwise when asked.
3. Pick your level of white support
In 2008, Obama had 43% support from white voters. In 2010, House Democrats had 38%. In 1984, Mondale had 35%. Polls are currently showing around 38-41% (splitting undecideds and leaving 1.5% other.) At least one pollster is known to have overestimated white support for Democrats in 2010, however.
Here's the tables:
Each Scenario from Step 1 has a table below. Use the table for your Scenario and your choices in Steps 2 and 3 to find Obama's final margin based on your assumptions.
If Obama can do just one point better among whites than Democrats did in 2010 - 39% - (which would occur even if preferences among whites stay the same as 2010, because of greater turnout from young voters) then Obama
cannot lose the popular vote as long as he maintains support among minorities at 2008 levels (Scenario 1) - and especially if he increases it (Scenario 2). It wouldn't even matter if turnout bucks history and yields an electorate 1% more white than the previous electorate. So the main question is - was 2010 a recent low point for white support for Democrats, or will it go lower? Details and explanations below the fold.
Demographic trends in the electorate.
At this point I think it's common knowledge that the electorate is growing less white over time, but what might not be clear is how robust that trend has been in the recent past:
This is no bouncy, uncertain wandering downwards. However, it is also not physics, and as such, the trend could conceivable reverse itself if racial minorities do not turn out to vote, or if the number of white voters skyrockets.
But let's look at a real-life worst-case scenario: 2010. I think we can all agree 2010 was a terrible year for Democrats. And what happened in 2010? The electorate was still one point less white than four years previously. If the demographic trend could withstand 2010, it is pretty safe to say it will continue in 2012.
So what do the data say this year?
We know the Latino citizen voting age population has grown by 22% since 2008, from Census data. We also know that in the past 20 years there has been a very strong correlation between the number of eligible voters and the number of actual voters. (See graph here.) We can use that information to predict a range of 10.4-11.8 million Latino voters this year, up from 9.7 million in 2008. If the number of all voters increases by 3% over 2008 (following population growth) then that would be an increase in the Latino share of the voting population of 0.5-1.9 percentage points.
We also have 'expectations' of another 600,000 new Asian voters this year, by way of the NAAS survey. That's an 18% increase that would bring the share from around 2.6% of the electorate to about 3.0%.
Are minorities registering?
But what about actual registration data? That is thin on the ground, but a quick search yields data for four states.
In Florida, since 2008, there's been an increase of registered voters of 28% for Asian voters, 10% for Black voters, 23% for Hispanic voters... and, coming nowhere close to balancing this out, a 2% increase in white voters.
In Georgia, it's 27% for Asian, 10% for Black, 34% for Hispanic, and 7% for white. And 76% for Other, which I wonder if it doesn't include a bunch of conservative 'post-racial' whites.
In North Carolina, it's 11% for Black, 70%(!) for Hispanic, and 3% for white. And 265% for Other, cripes! (Note that registration is ongoing in NC.)
In South Carolina, it's about 7% for both whites and non-whites, with a slight advantage for non-whites.
The share of non-white registered voters (not including the ever-popular Other) is increasing in all these states. There is no evidence of a surge of white registered voters. There is no evidence of a registration deficit for minorities, despite the concern trolling. (And not very good trolling at that - the premise of the article is OH NOES! Minority registration dropped 2%! Obama is doomed! But later, buried in paragraph 11 showing that, oh yeah, by the way, white registration also dropped - by 6%. Not to mention the numbers are two years old.)
Fine, but just because somebody's registered doesn't mean they'll vote. It could be that a ton of minorities who took the time to register won't bother to vote.
What about actual votes?
There's been plenty of analysis of early voting in swing states, but I worry that these numbers are unduly influenced by differences in the local ground game compared to 2008 and are not reflective of the country as a whole. (Although of course it's great for winning the electoral college.)
I looked at Cook County, Illinois early returns, and it seems that while the initial days of early voting are more popular than in 2008, the heaviest turnout is in suburbs where McCain won. Promotions and changes make it difficult to compare though. But chalk one up for the 'white conservatives are more likely to vote this year' theory.
Contradicting that idea is the Oregon ballot returns data. This is a nice apples-to-apples comparison to 2008, because the voting mechanism (vote by mail) hasn't changed, and Oregon is not a swing state this year, nor was it in 2008. Overall, a greater percent of ballots have been returned compared to 2008. Returns are about the same for both Democrats and Republicans, very similar to 2008. So in Oregon at least, there's no evidence of a massive Republican voter surge, nor of terribly dispirited Democrats. Update 11/2: Returns are now lagging 2008, and Republican turnout is one point ahead of Democratic turnout (still "about the same" but a little less so than before).
But minorities aren't enthusiastic! I heard it on NPR!
Utter poppycock.
First, let's look at African-American voters. 44% of Black citizens over 18 voted in 2010, the most in a midterm election in 20 years. If the African-American community managed to chalk up higher than normal turnout in the lousy year of 2010, why would turnout slump in 2012?
We also see the excitement of African-American and Latino voters building this year. In fact, in the past three weeks of Latino Decisions polling, 40-45% of Latino voters have said they are more enthusiastic than in 2008, with 20-24% saying they are just as enthusiastic. And, I might mention, 8% have already voted, and 87% say they are 100% certain to vote.
Obama's minority support.
So it looks like minority voters are registering and they're enthusiastic. But will they support Obama with the same gigantic margins they did in 2008?
Not likely. The margins will be even bigger, if polls are to be believed. Here's some apples-to-apples comparisons, which were used to create the table for Scenario 2:
African-Americans:
Pew. Fall 2008 O+88, fall 2012 O+89.
Asian:
NAAS. Fall 2008 O+15, Fall 2012 O+22.
Hispanic/Latino:
Pew. Fall 2008 O+36, Fall 2012, O+43
Gallup. Spring 2008 O+29, Spring 2012 O+41
Latino Decisions. Summer 2008 O+37, Summer 2012 O+48
Here's trends this year among Latino voters (note Romney's goal is 38%, or approximately a 22-point margin):
Appendix: The tables behind the tables.
The demographic breakdown for the tables can be seen
here, and the minority support tables for each Scenario can be seen
here. The 2008 and 2004 support levels are taken from exit polls; the other numbers are informed speculation on my part. The 2008 exit polls numbers were used as a base and adjusted from there. For instance, Latino Decisions generally has more Democratic-leaning numbers (and likely more correct numbers) than the exit polls, so I used the change in numbers from 2008 polls instead of the absolute value of the 2012 polls.
