What do the Exit Polls Tell Us About the Michigan Primary?

The results of the exit polls tell us about the impact that "crossover" voters had on the Republican primary, and give us a hint about what the Democratic race might have been like had the other candidates not been taken off the ballot. Check below the fold for a few quick notes.

Crossovers

According to the exits, the Republican primary vote consisted of 7% Democrats, 25% Independents, and 68% Republicans. Some of the pundits sounded surprised that a smaller proportion of Democrats and Independents voted in this primary, as opposed to the 2000 Republican primary where self-identified Republicans were less than 50% of the voters.

It shouldn't be that surprising. In 2000, we already had a presumptive nominee on the Democratic side. This time around, when faced with choosing a ballot, most of the party faithful were going to pick Democratic (out of habit, party loyalty, or pro/anti Hillary). That diminishes the pool of "very engaged" voters who would cross over to the Republican primary. Even though our race wasn't particularly exciting or meaningful, it was still a Democratic election.

So what impact did these crossovers have on the Republican primary? According to the exits, McCain took the plurality of Democrats (41%) and Independents (35%) who voted in the Republican primary, but didn't pick up nearly enough votes among those groups to close the gap between he and Romney among conservatives.

Doing some simple spreadsheet math (row percentages to column percentages), we can approximate the support that each candidate received by party:

McCain and Ron Paul greatly benefited from the presence of self-identified Independents (and to a lesser extent Democrats), whereas Romney, Huckabee, and Thompson did not.

Clearly, Kos' effort to get Democrats to propel Romney to victory didn't work. It was still a cool idea, though.

What could have been

The NEP exit survey asked respondents who they would have voted for if the rest of the candidates were on the ballot. Although this doesn't tell us what the race would have been like if the primary was still on Feb. 5, it does indicate that the race would have been much different than the outcome we saw yesterday.

This chart tells us that if the same universe of individuals who turned out on election day could have voted for any candidate instead of the choices that were available to them, 45% would vote for Clinton, 37% for Obama, and 12% for Edwards.

So 14% of Obama supporters and 27% of Edwards supporters went for Clinton instead of their preferred candidate. That alone reflects how meaningless the election was, since we know for sure that people who ordinarily would have voted for Edwards or Obama voted for Clinton instead. When interpreting the results, it isn't as simple as assuming that Uncommitted Vote = Obama supporters + Edwards supporters -- having a limited selection of candidates on the ballot absolutely made a difference in voting behavior.

Although we can use the results to asses the legitimacy of the election, we can't use this to determine what the result would have been had the candidates actually been on the ballot and allowed to campaign here. This is because the path to election day would have been different. Candidates and their surrogates would have actively campaigned here, and the narrative of the race would be different than repeatedly hearing about how Obama and Edwards turned their back on Michigan.

The outcome would also be different because a different universe of people would have turned out to vote than the group we saw today. For example, more Obama and Edwards voters might have turned out if the name of their favorite candidate been on the ballot. Voting for a person (especially someone that voters identify with) "feels" much different than voting for an abstract concept, and undoubtedly affected motivation to vote.

Other (pretty obvious) notes

* Among the 42% of the Republican primary voters who thought that Romney's ties to Michigan were important, Romney did very well. Still, we don't know if Romney's ties to Michigan caused him to perform well, or if voters decided on him first and then justified their choice with his ties to Michigan afterward. But the case can be made that Romney had a significant advantage here that won't carry over to other states.

* Romney does well with voters who are favorable toward the Bush administration, and McCain does well among voters who have a negative view of the Bush administration. Given how out of step the Republican base is with the rest of the country on their feelings about Bush, Romney's victory was probably good for Democrats overall. Even if he doesn't become the nominee, his continued presence in the race puts pressure on McCain to move to the right on issues where McCain disagrees with Bush. In this political environment, that will ultimately help us in the general election.

* McCain wins among those who disapprove of the war in Iraq (despite his pandering), whereas Romney wins among those who approve of the war in iraq. Similarly, McCain wins among voters who favor a path to citizenship, whereas Romney wins among voters who favor deportation. Again, Romney appears to win with voters who are in line with the Republican base, whereas McCain won with voters who weren't in line with the base on these issues. Given that McCain and Romney really aren't all that different on Iraq, this might be suggestive of each candidate's fundamental appeal to voters.

* Despite not being on the ballot, Obama still beat Clinton among African-Americans.

* A surprisingly large proportion of Edwards supporters (14%) broke for Kucinich and not Uncommitted, especially compared to the 3% Obama supporters who went for Kucinich. What does this tell us about each candidate's supporters?

* Of the different gender and marital status combinations, only single men broke for Uncommitted over Hillary.

Anyway, this is all speculation, very heavily informed by my own biases. What do you think?