The Undecideds

Yesterday, with the Case of the Puzzling Polls, I took a look at some polls’ lack of consistency and reasons why one might be skeptical of them. But that’s not the only factor that might make for volatile polling in the next six days until Iowa and two weeks until New Hampshire. We also must consider the undecided voter.

General election undecided voters have always puzzled me. What’s their deal? Can they really not make up their minds between two candidates whose parties’ ideologies are so often oppositional? (I’m assuming that’s the dilemma for most of them. If they’re like me and have little patience for the self-serving games played by both parties and instead ruminate on which third party candidate to support, then these many undecided voters would be casting a lot more protest votes like mine. General election results do not suggest that is happening, however.) Undecided voters in primaries, on the other hand, make a lot more sense. I can see the genuine conflict in the mind of a Democrat who identifies with Sanders’s true blue liberalism but considers Clinton the more electable nominee. Similarly, Republican voters, in order to decide between similar candidates, might need to parse through the constitutional approaches of Cruz and Paul, the moderate gubernatorial records of Christie, Kasich, and Bush, or the laughable statements of Trump and Carson.

How many undecided voters are there? We can never really know. Pollsters survey what candidate people are leaning toward, which keeps their undecided figures extremely low, but people who lean one way in the beginning of January might tilt elsewhere by the end of it. I would still categorize these voters as “undecided.” To best guess what percentage of the electorate might be blowing in this breeze, we should look to recent history.

It seems that undecided voters usually make up their minds in the final days before casting their ballot. I’ll yet again dip into FiveThirtyEight‘s vault to dig out a better chart than I could ever make:

On average, nearly 40 percent of Iowans and half of New Hampshirites make up their mind in the week before the election.

This indecision creates serious practical implications for us in the prognostication business. For the sake of round numbers, what if we say that 40 percent of Iowa Republicans (the above chart suggests 43 percent and trending up) are either undecided, or they’ve told pollsters they’re leaning toward one candidate but will change their mind in the last week. If that’s the case, here are candidates’ Real Clear Politics averages in Iowa followed by the percentages of Republicans that are actually committed to them:

1. Trump 32—-19
2. Cruz 27——-16
3. Rubio 12——-7
4. Carson 7——-4
5. Bush 4———2
6. Paul 3———-2
7. Christie 3——2
8. Huckabee 2—–1
9. Kasich 2——-1
10. Fiorina 2——1
11. Santorum 1—-1
12. Gilmore 0—–One voter becomes four-tenths of one?
13. Undecided—-40

In New Hampshire, where half of voters seem to make up their mind in the last week (the higher slice probably stems from the Iowa results affecting their decision), the numbers flatten even more:

1. Trump 32—–16
2. Cruz 13——-7
3. Kasich 12—–6
4. Rubio 11——6
5. Bush 8——-4
6. Christie 7—–4
7. Paul 4——-2
8. Fiorina 4—–2
9. Carson 3——1.5
10. Huckabee 1—-0.5
11. Santorum 0.2–0.1
12. Gilmore 0—–Some negative number
13. Undecided—50

If we look at polls through this undecidoscope (what do you mean that isn’t a word, WordPress?!), we see a much tighter and more volatile race.

The problem with volatility, however, is that it’s hard to predict. That’s what makes volatile things volatile. How will these numerous undecided voters apportion? Consider these three groups of predictions:

• One school of thought might be that these undecided voters will eventually get divided proportionally, bringing us back to the poll numbers to which we’ve grown accustomed.
• Trump supporters will point to his rabid fan base being much less likely to desert him, and that not only will he lose almost none of his Iowa and New Hampshire, but when people do decide or change their minds in the last week, they’ll actually add to his total share.
• Those who are rooting for Trump’s demise — chiefly his Republican rivals, people with a college degree, and the nonpareil Presidential Politics for America — will instead reason that anyone who will end up supporting Trump has already done so.  In other words, the undecideds are not vacillating between Donald Trump and John Kasich.

That third scenario is not impossible. Trump is notoriously divisive. I’ve mentioned on many occasions that he struggles with “second choice” numbers and his favorability among Republicans is generally lower than all his top rivals except for Bush. As a result, few Republicans are likely to move to him in the final week. For the undecideds, they’re not waffling between Trump and someone else. Their dilemmas right not are between the likes of Cruz and Rubio or Kasich and Christie or Gilmore and a ham sandwich. I don’t even think many voters are torn between Trump and Cruz. Cruz supporters are either deeply conservative (which Trump is not, but Rubio is), deeply constitutional (which Trump is not, but Paul is), or deeply evangelical (which Trump is not, but Carson is). So even if someone is considering Cruz, Trump is probably not their other option.

Between our potentially untrustworthy polling and this undecided factor, we must be prepared (and hopeful!) for a chaotic couple of weeks.

Six days until we start to see just how accurate these polls are! That’s almost as exciting as the actual race.