Political science and election science, as social sciences, are sandwiched between the desire for mathematical exactness and methodological limitations to measuring human behaviour. In other words: political scientists, much like economists, want to give precise answers by relying on data but people’s preferences for candidates and their political points of view are difficult to measure. At the same time, both the ever-increasing availability of data and the current Zeitgeist make elections more and more data-driven exercises. This brings us back to a pivotal question of the political scientist: what is central to our discipline? Data? Or human insights? Quantitative or qualitative?
Cognitive dissonance
A couple of stories on the Bush candidacy this week highlighted this methodological rift and the different conclusions one can arrive at by either looking at the data or by relying on historical insight. There seems to be a cognitive dissonance between the fact that Trump has been the lone and distant leader of the GOP polls for a good three months now and the refusal to believe he is actually likely to win the nomination. As this article in the Washington Post puts it: Trump’s popularity still seems temporal.
Let’s look at what the numbers have to say. The article in the Post seems to suggest, albeit skeptically, that the data tell a story of a nearly inevitable Trump victory in the primaries. Whereas Romney, the eventual GOP candidate in 2012, led in the polls on and off for short periods at a time, Trump’s lead in the polls has been much more consistent. The author, despite the data evidence, is still skeptical about a Trump victory because Romney took the lead at the right time: he rose to the polls just before voting actually started in February. At this point, we are still quite some time away from the first primaries in Iowa and New Hampshire.
Donor data
This article in the conservative magazine National Review puts it more bluntly: Jeb Bush is ‘toast’. The (anonymous) author gets to this conclusion by look at, among other things, fundraising data. The central premise is that an electable candidate needs to appeal to both big donors as well as the rank-and-file members of the party, who may chip in $100 or so. Popularity with big donors not only brings in the big bucks but also signifies establishment backing. (Mitt Romney was an example in 2012.) Grassroots support is equally necessary to win a general election. (Again, Mitt Romney is an example of a candidate whose grassroots support was tepid at best.) According to these data, Jeb Bush’s chances are slim to none-owing to a ‘near-complete lack of support’ among small-time donors (up to $200). According to this model, Ted Cruz would actually be most likely to win the nomination. Ouch.
Bush on the ticket: 5 out of 6
One can also look at history, however, as this article in Politico did, and arrive at the opposite conclusion. The author argues that, as I wrote here earlier, primary voters will eventually want go with the candidate most likely to prevail in the general election. This is a middle of the road candidate, more often than not a little boring but a safe choice nonetheless. This especially the case with a conservative party, such as the GOP, which is likely to make, well, safe and conservative choices. The last time the GOP went with exciting and extreme over safe and moderate was when it nominated Barry Goldwater in 1964, who then lost with a landslide to Lyndon Baines Johnson. Ever since then, it went into the general election cautiously. Sometimes this ended with a dud, such as with Mitt Romney. It is telling, however, that the GOP won five out of six elections when there was Bush on the ticket. (Then again, the only time a Bush lost a presidential ticket was against a Clinton, in 1992.)
Gamblers
An interesting but often overlooked source of information are the bookies. An economics editor at Dutch NRC Handelsblad wrote a story about betting on presidential candidates, as if it were a horse race. According to the sources he used, Predictic and Pivit, Clinton had the highest change of winning the general election, with 49%. Trump’s chances at 18% are significant but not nearly front running. Bush chances of winning the Republican primary are 28%, while Rubio is cast as the frontrunner with 40%.
Call me a nostalgic romantic sentimentalist, but I’ll go with history over the data. (Or maybe just call me a historian …) The data revolution is definitely useful for analysing political landscapes and efficiently reaching electorates in particular. Nevertheless, there have been ample examples of grave polling errors in recent elections and methodological problems remain.
Jan van Ewijk's personal website 
What the polls do not show is the second preference people hold. Seeing there are still so many candidates in the race, which will likely be reduced before and during the primaries, the question is where former supporters of ex-candidates will go to. I think the odds are high that many people will stick with Trump, but that the majority of the people who did not support Trump from the start, will vote for anyone who is not Trump and leading in the polls, which will in part be determined by the supporters of the candidates that quit the race first.
This might be extremely random, just like NDP lost to the Liberals in Canada, purely because they were the smallest of the two main opposition party. They would likely have been much larger, had they stayed ahead of the Liberals until election day.
So do you have any information on second and third preferences?
Hi Joes, thanks for taking the time to write a comment. Yes, second preferences matter a big deal and I hadn’t looked at that closely. What possibly also matters a great deal is the preference of GOP and Independent voters for the GOP candidate versus Clinton, who I don’t imagine losing the nomination after last week. The two gambling sites I mentioned also sell ‘contracts’ on those preferences. As for your question, I don’t actually have those myself but I am sure you can easily find them on the web, e.g. RealClearPolitics. Strategic voting in the primaries is definitely a factor: presumably voters will come down to the sensible moderate after making some noise in the beginning of the primaries, to voice their concern, before voting has even begun. I would think this dynamic will favour Bush, but he’s in serious trouble after last night’s debate. So, the real questions, then, is who will be the next best moderate, establishment candidate for the GOP?
I had not thought of the influence of the Clinton vs Republican candidate polls yet. That could of course also have a major impact on which moderate it will be, given that he/she is the one who attracts most swing voters. I doubt that will be Bush, I think Clinton would beat him easily, so indeed, who will it be?
Bush was about the only moderate in this GOP field. Rubio seems to be the guy if you look at websites such as Predictiv and Privit. I can’t see Carson being a serious candidate for a general election, Cruz is too right wing to attract independents and swing voters, not too even mention Trump, of course. Seems like they have really shot themselves in the foot by not having a candidate who can beat the Clinton machine.