About this Blog
About The Author
Email Me

RealClearPolitics HorseRaceBlog

By Jay Cost

« Are the Chickens Coming Home to Roost? | HorseRaceBlog Home Page | The Puerto Rico Wild Card »

Overworking the Primary Data

In the many discussions of the Democratic nomination on this page - we have talked about the fact that Hillary Clinton and Barack Obama must make arguments (or, as Bob Shrum has said, moral claims) for the nomination. Such claims will have two general points - legitimacy and electability. Each candidate will argue why he or she is the true choice of Democrats, and why he or she is more likely to prevail in November.

We have not heard a lot of squawking from Clinton about legitimacy just yet. Part of this has to do with the fact that, by any reasonable metric, Clinton is in second place. Be it pledged delegates or the vote count - Obama has at least a slight lead. If the primary process were to end today, he would likely be seen as the legitimate nominee. This precludes Clinton from making a claim to legitimacy. She will be able to make a claim for legitimacy if she pulls ahead in votes, but until then it behooves her to avoid discussing legitimacy altogether. Why establish a rhetorical standard that she might not be able to meet?

So, much of the back-and-forth between Clinton and Obama's flacks has involved electability. Unsurprisingly, we have seen both sides making claims that do not stand up to strict scrutiny. For instance, a reader writes in with the following question:

I've been hearing a lot lately that since Hillary won battleground states of OH, FL, and MI (and will probably win Pennsylvania) then she is better positioned to win those states in the general election. But is this actually true?

Not necessarily. Clinton's flacks can indeed be seen on the airwaves arguing this point - but in so doing they are committing an inferential error. What they are assuming is that because partisan Democrats (her core support group) in a given state support Clinton over Obama - the entire state will. This need not be the case. It could just as easily be that Independents and persuadable Republicans would prefer Obama to Clinton in those swing states. So, in an ironic twist, Clinton would win the primary but not the general. Perhaps the Clinton campaign wishes to argue that Obama could not win in the general the voters she has won in the primary. Maybe - but the primaries alone do not indicate whether that is the case.

Obama's supporters have made their own errors. For instance, one can often find them arguing that his primary strength with self-identified Republicans is evidence of an advantage in the general. By themselves, primary results cannot indicate this. The only data we have on these voters is their self-identification. We do not have histories of how they voted in general elections. To argue that Obama's margins among self-identified Republicans is a sign of strength in the general, they would have to show that these voters are typically reliable votes for Republicans in the general who are being wooed away from their party. We cannot assume that this is the case. Remember that there is always a portion of each party that votes for the other side. So, these voters might actually be reliable Democratic supporters who see themselves as Republicans (lots of people see themselves that way..."I vote for the person, not the party" - but it always seems to be that the better person is of the same party!). In that case, their support for Obama does not necessarily portend general election strength. [Though I think Obama is on better ground to point out that, according to the opinion polls, Independent voters seem to like him more than they like Clinton. Likability is a major prerequisite for vote choice - and so this might be a sign that Obama is stronger than she among this group. At the same time, the big question - how much stronger is he? - cannot really be answered in polls conducted eight months before the election.]

Take an example of the difficulty in using the primary returns to argue about strength with Republicans. Last month in Wisconsin, 9% of all Democratic primary voters claimed to be Republican. That works out to be about 100,000 voters. Obama won about 70,000 of them. In the 2004 general election, John Kerry won 8% of self-identified Republicans, according to the exit polls. That implies about 240,000 self-identified Republican voters who supported Kerry - roughly 170,000 more than Obama won last month. So, is Obama expanding the base of the party to voters who typically vote Republican, or is he "merely" enthusing self-identified Republicans who tend to support Democrats? The primary data alone cannot answer this question.

Now - here's the caveat. The point is not that these arguments are untrue. It is that they are underdetermined, which is another way of saying that we just don't know either way. Personally, my intuition is that both might be correct. But that's all I have right now - an intuition. These theories are plausible, and worthy of testing (I am working on a way to test the Clinton camp's argument right now) - but that's it.

Nevertheless, we shouldn't be too tough on the Clinton and Obama flacks for overworking the primary data. Sometimes, you have to throw a lot of arguments against the wall to get a few to stick. And anyway, the primary results have been overworked by lots of people. For instance, some Democrats - and many in the media, for that matter - are pointing to the relatively high turnout in the Democratic primaries as evidence of an enthusiasm gap that advantages the Democrats. Without commenting on who has an advantage in November, I will say that this particular argument is problematic.

Roughly 62.2% of all primary votes have been cast in the Democratic primary. This is an impressive statistic. However, by itself it does not count as evidence of a Democratic advantage. The reason is that Democrats typically out-perform Republicans in the primaries. The following chart compares the Democrats' share of primary turnout against their share of the two-party vote in the general election.

Primary Versus General.jpg

As you can see - 62.2% is far from extraordinary. Even when we exempt the years in which the Republican Party had non-competitive contests (1972, 1984 and 2004), the Democrats typically out-perform the GOP. Pulling in 62.2% of the primary vote is no unique feat for the Democrats. 1996 is telling. Bill Clinton had no serious challenge while Bob Dole faced a protracted battle against multiple opponents. And yet the GOP still only pulled in 55% of the primary vote.

Another key year is 1988. This is the best apples-to-apples comparison of 2008 that there is. That year, both parties had open nomination battles. The Democrats out-performed the GOP by a margin larger than what they have done this year, pulling in a little more than 65% of the total primary vote. Did it do them any good in the general? No. George H.W. Bush beat Michael Dukakis, 54% to 46%.

So, there is apparently no connection between Democratic primary turnout and the Democratic vote in the general. Why not? I would suggest two reasons. First, higher turnout is in many respects a consequence of drama rather than enthusiasm. In years past, the Democrats have had more dramatic primary battles that have intrigued and engaged voters. This year is no exception. Dramatic races might actually have a negative effect on the party because it drains time and money from the eventual nominee.

Second, our system does not weigh votes according to enthusiasm. I think it is clear that there is an enthusiasm gap this year. I also think that part of the vote difference between the GOP and the Democrats might be a consequence of this gap. However, enthusiasm can only do so much for a candidate. If Voter A can't decide whom to support on Election Day, and breaks the tie by flipping a coin - his vote counts as exactly one vote. If Voter B is so excited to support his beloved candidate that he can't sleep the night before - his vote counts for exactly...one vote!

Where enthusiasm has an effect is in the relative likelihood that Voters A and B will vote. Voter B is almost assured to vote while Voter A is much less likely. This is the benefit that accrues to the candidate with enthusiasm on his side. However, the likelihood of Voter A actually voting increases as the competitiveness of the contest increases. Competitive elections generate attention and interest, and therefore participation. This is why, for instance, turnout was down 8 million votes between 1992 and 1996, was back up in 2000 and even higher in 2004. What's the difference? 1992, 2000 and 2004 were intense elections that captivated the nation. 1996 was not.

Where I think enthusiasm could have the biggest effect is in the money both candidates are able to bring in. This is probably one reason McCain is going to press Obama to take public financing. Obama's enthusiastic voters count for just one vote apiece, but they can give him a financial edge over McCain that public financing would nullify.

-Jay Cost