Tonight Nate Silver published a very interesting analysis about whether or not Mitt Romney could beat President Obama using a various combinations of the ethnic turnout and electoral choice compositions of both the 2004 and 2008 elections.
There are a few problems with his analysis in my mind. He - only somewhat, as you'll see - assumes, and correctly I think, that the two biggest questions of this election are the choice behavior of whites and the turnout levels of minorities. Instead of going in depth, however, Nate simply relies on the numbers from 2004 and 2008 and intermixes these in three hypothetical scenarios.
This diary has been posted to DK Elections, an official Daily Kos sub-site. Please read the DKE Mission Statement. Our focus is on electoral politics rather than policy or preference. Welcome aboard!
The first being that turnout remains the same from 2008, but that the choices of each ethnic group revert back to their 2004 breakdowns. This presents a real problem. Does anyone think, realistically, that Romney will reprise Bush's standing with any minority? In no single poll conducted does Romney even come close to breaking 35% among any minority. Under this scenario, Romney ekes out a very very slim 49.5-49.4% win. Electoral college aside, I'd count this as a Romney victory in such an analysis.
The second hypothetical is that turnout reverts all the way back to the 2004 breakdown, but that the choices of each ethnic category remain the same as 2008. This presents an even bigger problem. Does anyone truly believe that the long-term trend of increased participation of minorities will abate, let alone revert back to 2004? I can buy an arguments for some level of slide of minority turnout, but not so large that composition mirrors 2004. Furthermore, does anyone really believe that Obama will be performing at the same levels among whites that he did in 2008?
The third hypothetical is a more nuanced take on the 2004 turnout - which, again, is a stretch - by giving Romney Bush's performance among whites. Obama wins these last two hypotheticals with a majority of the vote.
Given that the two biggest questions of the election are minority turnout and Obama's percentage of the white vote, I lay out an additional five reasonably plausible scenarios which all use these baseline assumptions:
- Romney will get Bush's share of the white two-party vote.
- Obama will reprise his share of the African American two-party vote.
- Obama and Romney will average the 2004 and 2008 breakdowns of Hispanic, Asian, and Other in the two-party vote.
Obviously I do all of my analysis in two-party vote share only, something that Nate neglects to do (which, to be honest, drives me nuts to no end oh my God in Heaven). Third party vote share does not matter in analyses like these. The only thing that matters about third party vote share is its
distribution. For instance, if the third party vote share is relatively higher in Virginia, it could have some effect on who wins that state (wink wink cough cough nudge nudge), but we should only consider it in that case because it could sway the outcome of the
electoral college, not the national popular vote. Third party vote share simply tells us nothing about Obama vis-a-vis Romney.
Scenario 1: 2004 Turnout
In this scenario, the ethnic turnout patters of 2004 are duplicated. Obama does better than does Kerry with Hispanics, Asians, and Others, but does not match his 2008 levels of support. He matches Kerry among whites and matches his 2008 support among African Americans. I happen to agree with the punditry that the biggest question of this cycle is how far above Bush's performance among whites that Romney would have to go in order to beat Obama. He'd have to achieve Reagan-esque levels by taking 58.31% of the two party vote to secure 50.01%, and probably go up to 59% in order to be out of the woods as far as the winning the electoral college as well (when you're in that grey area between 49.5% and 50.5%, winning the electoral college as well isn't guaranteed). Factoring in third party vote share, that'd mean that Romney has to win anywhere between 60 and 62 percent of the white vote in order to be President. This, by the way, is the worst case scenario for the President and really underscores how hard a time Mitt is going to have winning the presidency.
Scenario 2: Balance
In this scenario, I've average the ethnic category turnout of 2004 and 2008. I think there's a good argument to be made that the minority turnout share of 2008 was somewhat of a high-water mark and that it will recede. Nonetheless, I think that receding to 2004 levels is extremely unlikely. Here we have whites moving back up to 75.15, with African Americans taking the largest hit among turnout (which I think presents its own problems). In order for Romney to get to that magical 50.01% he'd have to take 59.28% of the white vote.
Scenario 3: 2008
In this scenario, the turnout breakdown of 2008 is replicated. Ofcourse this might be a bit of a stretch, but you never know. I think for any discussion to be full, you have to include the bright spots for both sides. Unfortunately, Silver's analysis only focused on the bright sides for Romney (well, brighter not bright, given that even under Silver's hypotheticals Romney would have a tough go of it). For Romney to win here, he'd have to bump his white share up to 60.3%.
Scenario 4: Minority Boost
In this scenario, I've done something which I believe isn't out of the question. African American turnout isn't going to be higher, but that doesn't mean that Hispanics and Other (which have been on the long-time increase in this country) won't increase somewhat as well. I don't expect the turnout boost, if there is one, to be anywhere near as high as between 2004 and 2008, so I've only increased the share in percentage terms by half the rate as between those two elections from 2008 to this 2012 model. Asians, obviously, don't increase at all as they made up the same level in both 2004 and in 2008. I actually think this might be somewhat of a wild-card. Asian immigration and citizenship levels have been rising much faster in recent years, so we may end up seeing a boost here as well. I'm not going to bank on it however... For Romney to win under these assumptions, he'd have to take 60.63% of the white vote.
Scenario 5: Complicated Picture
This is actually what I consider the most likely outcome. It takes the increase of other minorities from the previous model, but actually rolls back the turnout levels of African Americans to the average of 2004 and 2008 in favor of a slightly higher white share than the Minority Boost scenario. I also think this explains very well the Romney campaign's strategies. They'd only have to win 58.9% of the white vote (very similar to the 2004 scenario) in order to win the election, which is certainly within reach. But they wouldn't have to do that at all (admittedly a tough haul) if they could improve on the margins of the overall minority vote. They've already technically done that simply because of the turnout picture among minorities only. African Americans make up less share than 2008, while the more Republican minorities make up a larger share.
This actually explains why they made such a huge push for Hispanics during their convention. If they were expecting a voter universe that was more like 2004, they wouldn't have done that at all because taking into account the overall minority share in a cost-benefit analysis would have had more costs. Instead they'd have devoted all their energy to bumping up that white share (which they'd need slightly less of). If they're expecting more Hispanics and Asians (or Other) than 2008 (not just 2004), the cost-benefit analysis suddenly changes to require that they pursue whites and minorities both so as not to put all their eggs in one basket.
Nonetheless, all of these scenarios underscore why it is so difficult for Romney to win. He has to achieve Reagan-esque levels among white voters as a candidate who is anything but. He's boring, stale, obscenely right, and has no common touch.