Nevada Republicans have finally finished counting caucus ballots, and Mitt Romney was declared the winner there with just slightly more than 50 percent of the vote. Newt Gingrich finished second with 21 percent, followed closely by Ron Paul at 19 percent. Rick Santorum was fourth with 10 percent of the vote.
There are three interesting data-driven takeaways from the state.
Maybe polling caucuses isn’t hopeless. As it turned out, the Nevada results weren’t that far from our forecasts, which I had been nervous about because they were based on only two polls and because Nevada polling has been erratic in the past. Mr. Gingrich did a few points worse than his forecast and Mr. Paul a few points better, but both were within the confidence interval established by the model. And Mr. Romney’s 50 percent of the vote was very close to the forecast of 51 percent.
The Republican race may be more stable than it appears. Sean Trende, at Real Clear Politics, makes a much more detailed version of this argument, but Mr. Romney’s 29-point victory in Nevada isn’t that hard to reconcile with his narrower win in Florida or other early-voting states.
Much of Mr. Romney’s margin of victory in Nevada was because of the presence of Mormon voters. According to the entrance poll of the state, 26 percent of caucus voters were Mormon, and 90 percent of those voted for Mr. Romney.
The entrance poll did not provide a specific breakout for non-Mormon voters, but we can derive one by taking its numbers on the Mormon vote and then estimating the non-Mormon vote such that the two groups together match the actual result in Nevada.
That estimate suggests that Mr. Romney won about 36 percent of the non-Mormon vote, putting him 8 points ahead of Mr. Gingrich, who won 28 percent.
This estimate isn’t perfect. There is sampling error in the entrance poll results. In addition, it was calibrated based on the assumption that Mr. Romney had won about 54 percent of the vote rather than his actual total of 50 percent.
Still, Mr. Romney’s 8-percent margin among non-Mormon voters is interesting because it roughly matches Mr. Romney’s advantage — around 9 points — in the total popular vote through the first four early-voting states, none of which have substantial Mormon populations.
This represents a mix of good news and bad news for Mr. Romney. On the one hand, if Mr. Romney has something like a 9-point advantage nationally over Mr. Gingrich, that margin should be more than enough for him to win the Republican nomination. With that kind of edge, Mr. Romney would lose some states where the demographics are less favorable to him than in Nevada, but he would win enough others that he should have the nomination wrapped up well in advance of the convention. (As I wrote yesterday, I tend to view the biggest threat to Mr. Romney as being a new dynamic in which Mr. Santorum rather than Mr. Gingrich becomes his main challenger.)
On the other hand, we do not have much evidence yet that Mr. Romney’s organizational strengths will translate into an automatic advantage for him in the caucus states. If the results in Nevada and the other states can be explained reasonably well by demographic modeling, that implies that whether the state holds a primary or caucus may not be all that pertinent. Many of the states that hold caucuses happen to have good demographics for Mr. Romney anyway, but we shouldn’t confuse the cause and effect. We will have considerably more data to study on this question, of course, after Colorado and Minnesota hold caucuses on Tuesday.
The better Mr. Romney does, the lower the turnout. This dynamic, which we highlighted after Florida, was in full effect in Nevada. Turnout there declined to 32,963 voters this year from 44,324 voters in 2008 — a 26 percent drop. And the decline was slightly larger, 29 percent, among voters who identified as Republican in the entrance poll.
So far, there has been a clear inverse relationship between how well Mr. Romney performed in a state and the strength of the turnout.
It does not appear likely that this is just a fluke; Michael P. McDonald, of George Mason University, has found that the relationship also exists if you look at turnout on a county-by-county level.
But is it likely to matter in November? Mr. Trende writes that the importance of the phenomenon might be overblown. He notes, for instance, that there is not a clear relationship between turnout in past primaries and turnout in the general elections in those years.
My view is different, although in a subtle way. Turnout data is tricky to work with. Ideally, you’d want to adjust for things like how competitive the nomination race was and for how long, how many states held caucuses instead of primaries, the peculiarities of the nomination calendar and the dynamics of two parties holding primaries or caucuses in the same year, perhaps along with other factors. There are also a lot of things you’d need to unwind in order to measure how strong turnout was in November.
And, as is the case with other types of election studies, the analysis is confined by a small sample size — only 10 presidential years from when the current nomination process was adopted in 1972 to 2008. This presents a significant challenge even when you are dealing with relatively well-defined variables, like a president’s approval rating. But it is even more of a problem when you’re studying something that is tricky to measure, like turnout.
When the empirical evidence is this ambiguous, it obviously does not validate the hypothesis you are trying to study — but it also does not necessarily refute it. There might be a modest effect there that is hidden by all the noise in the data, or there might not be.
It should be remembered, however, that Republicans usually have the turnout advantage in November because their voters tend to come from demographic groups (like older Americans and wealthier Americans) who vote more frequently. This usually manifests itself in the fact that polls of likely voters show somewhat more favorable results for Republicans than polls of registered voters.
The safest default assumption is probably that this gap will exist again, but that it will amount to a more typical value like 2 or 3 percentage points than the 6-point “enthusiasm gap” that existed in 2010.
Or it could be that the middling enthusiasm for Mr. Romney will only make much difference if he appears to be in trouble by November. Democratic turnout was quite poor, for instance, in 1984 for Walter Mondale, a candidate who has some parallels to Mr. Romney. However, it was clear that Ronald Reagan was going to win that election anyway; low Democratic turnout contributed to Reagan’s margin of victory, but strong turnout would not have reversed the result.
On the other hand, Democrats had somewhat limited enthusiasm for John Kerry in 2004 — but that election was much closer, and they did not have any major problems in getting their voters to the polls.
In other words, perhaps if Mr. Obama appears poised for a 6- or 7-point victory by November based on the economic fundamentals, Republican voters may feel that their vote makes little difference anyway and some of them will stay home as a protest, expanding Mr. Obama’s victory margin to 8 or 9 points instead and making it look prettier in the Electoral College. But I’m more skeptical that this will matter much in an environment in which the election will be very close and every vote could make a difference.