It looks like it’s about time to start modifying my forecasts on the basis of likely candidates. Currently, I think that HRC (0.91) is much more likely than Bernie (0.09) to win the Democratic nomination, while Trump (0.47) is running ahead of Rubio (0.32), Cruz (0.18), and Kasich (0.03). The current outcomes of the polls on head-to-head matches are as follows.
- HRC vs. Trump: 0.43 likely to happen, HRC leads Trump by 3%.
- HRC vs. Rubio: 0.29 likely to happen, Rubio leads HRC by 5%.
- HRC vs. Cruz: 0.16 likely to happen, Cruz leads Clinton by 1%.
- HRC vs. Kasich: less than 0.03 likely to happen, so I’m ignoring it for now.
- Bernie vs. Trump: 0.04 likely to happen, Bernie leads Trump by 6%.
- Bernie vs. Rubio: 0.03 likely to happen, Bernie and Rubio are in a dead heat, so I’ll give a 0.015 chance of victory to each candidate.
- Bernie vs. Cruz: less than 0.03 likely to happen, so I’m ignoring it for now.
- Bernie vs. Kasich: Not in this lifetime.
So the Democrats win in 50.5% of the cases, and the Republicans win in 46.5% of the cases, with the remaining 4% divided between them. So the likelihood of a Democratic win is 0.525 and that of a Republican is 0.485, right?
Well, there are at least (at least!) 3 reasons to want to slow down a bit. First, the leads in the head-to-head matches are not overwhelming. Even if, e.g., HRC leads Trump by 3%, an HRC loss is well within the realm of possibility. Second, it’s far too early for these polls to have a lot of predictive value. Third, these numbers don’t take into account long term election trends, e.g., the rarity of a single party holding to White House for more than 2 terms (as mentioned by @morrelldaniels).
Now, my original prior (about 0.45 Democrats and 0.55 Republicans) was based on the frequency of party wins since the mid-19th century, and that fact probably (I hope) means that I’ve already incorporated (in an admittedly very crude way) long term trends like party-switching. So I’ll stick with the practice of modifying my prior, rather than starting from scratch. And, even if the numbers I have from the head-to-head matches is not highly predictive, it’s likely to have some predictive value now. And if I keep updating, it’s likely to have even more predictive value in the future. In sum, very small advantage (over my previous forecast) to the Democrats.
- 50% A Democrat
- 50% A Republican