7 Types of Undecided Voters

 

Yes, Undecided Voters Exist.  But Who Are They?

The election is over, and now everyone can weigh in about who did a good job, who did a bad job, and why.  One group of people that is coming under an enormous amount of fire is the pollsters.  For instance, Nate Cohn of the New York Times has written a piece about how many of the polls were once again off - especially by underestimating the size of Donald Trump's support.  Cohn is not alone.  Many people, especially on twitter, have gone ballistic against our nation's predictors and prognosticators, concentrating a lot of their artillery on Nate Silver of the 538 website and podcast.

To which Nate Silver replies, (and I quote) "Fuck you, we did a good job."  Nate Silver's response to all such criticisms is always that he is a data analyst and not a pollster, and that if the pollsters are not doing a good job, it's not his fault.  Very well, Nate, we're not trying to blame you, but if you're a data analyst, then, well, analyze this data.  What went wrong?  Why are pollsters perpetually getting their predictions wrong?  

And before we go any further, let's throw out one pet hypothesis: the "shy Trump voter".  That is, people who vote for Trump, but who are shy about telling pollsters they will do this beforehand, because Trump is personally so disliked.  I'm sure such people exist, but I'm not sure they're all that statistically significant.  After all, it seems that the pollsters got other races even more wrong than they got the presidential race.  Downballot conservatives did even better, contrary to pollster's predictions, than Trump did.  So, if that's not the reason the polls were off, then what is?

Well, I won't claim that I know the answers.  But look at it this way: in order to construct and carry out a poll, pollsters (and data analysts) have to have some assumptions and some working hypotheses about the people that they are polling.  But for many people, the choice between Trump and Biden was so stark, that it's hard to imagine the mindset of the "undecided voter".  And if you cannot imagine the mindset of a person, it's going to be very difficult to construct a poll that accurately captures their opinions.  

Sometimes this is downright funny.  Comedians on twitter, like Michael Ian Black, were making fun of the very idea of an undecided voter.  How could anyone be undecided in an election like this?  So we get sarcastic tweets along these lines:

But I think part of the problems of pollsters - and the rest of us - may have something to do with a failure of imagination. It's unimaginable to us that someone would think along the lines Michael Ian Black is describing, and not without reason.

One thing that drove me crazy during the 2020 election cycle (which started... uh... in late 2016?) was the condescending attitude and terrible political strategy of many Democrats.  I'm thinking especially of Jon Favreau, host of the "Pod Save America" podcast, who created a special, side-project podcast called "The Wilderness," which concerned focus groups that he and his colleagues interviewed as election research in several swing states.  I listened to every single episode of "The Wilderness" even though it made me want to tear my ears off.  What made me insane was the way in which Jon Favreau and his colleagues would ask "undecided voters" questions, and then not listen to those people's answers.  No matter what the respondents said, Favreau and his crew just would not hear them.  They had a prefabricated idea of what an "undecided voter" is, and tried to shoehorn the people they were interviewing into that ridiculous stereotype.

The big problem, as I see it, is that terms like "undecided voter" and "swing voter" are ambiguous, and our pollsters, analysts, data scientists and statisticians are not doing a good enough job distinguishing clearly between these various subgroups within the larger headings.

So let's work a little on our imagination and our empathy.  Let's take a moment to review the various kinds of "undecided voters".  For the purposes of this essay, I'll refer to the Biden/Trump election, but much of this probably applies, mutatis mutandis, to all kinds of elections, now and for the foreseeable future.  In 2020, who were the undecided voters?

1) People who like both Trump and Biden, but couldn't decide which one they liked better.  This number, I would estimate, is so vanishingly small that it may be difficult to measure.  The percentage may fall within the margin of error of zero.  (...but maybe I'm wrong!)  The main problem with pollsters is that when they try to imagine an undecided voter, they imagine someone like this, and format their questionnaires accordingly.  But.... do you know anyone like this?  I don't know anyone like this.

I would guess that most of the people who were "undecided voters" were people who dislike both Trump and Biden, and quite often people who hate both Trump and Biden, or who are, at best, indifferent to one or both of them, but cannot make up their minds for other reasons.  These can be broken down into smaller groups:

2) People who are further to the left than Biden. Some of these people felt betrayed by the Democratic Party, and may resent the influence the Democratic Party has in the media and other institutions, constantly hectoring them to vote a certain way.  So they often had a lot of difficulty bringing themselves to vote for Biden, and may have considered voting for a third party or not voting, but also didn't want Trump.  This left them unsure about what to do.

3) People who are further to the right than Trump.  This might include real conservatives, culturally traditional people who despise Trump's pride, vulgarity and lack of respect or sense of duty towards the institutions in which he holds office, and who therefore had difficulty bringing themselves to vote for him, but who are afraid of the changes that might come with a Democratic administration.  Or it may include libertarians, who once again considered voting third party.  It may even include white supremacists, Nazis, and conspiracy theorists who think Trump is too compromised.

Groups (2) and (3) may have been undecided for quite rational reasons.  They may have been uncertain what is the best strategy to push (what could be) their own party further in the direction they'd like to see it go.  It's a genuinely difficult strategic question.

4) People who were (and are) completely disgusted with the entire enterprise of American government.  Such people say things like "They're all crooks" and won't support any politician.

5) People way outside the familiar political spectrum, with all kinds of beliefs that don't fit neatly in the "left" or the "right".  Some religious beliefs might come into play here.  I once met a person who was a militant vegan, who determined his political opinions by trying to find out the dietary habits of politicians and figuring out how many animal products they consume.  Some people may believe the U.S. government is controlled by space aliens, or that we're living in a simulation.  There are all kinds of ways of looking at the universe out there, and it's not surprising that many - probably most - of them are not represented by the two major political parties.

6) People who don't have any political opinions.  These people don't follow the news, and are not sure what is going in the world today.  We are taught to look down on such people, but have some empathy - they may be busy caring for 10 children, or for a sick family member.  Or perhaps they are idiot savant geniuses who occupy nearly 100% of their thoughts trying to solve obscure mathematical theorems, and often forget to eat or bathe.  Or maybe there are people who don't speak English, and haven't found reliable media about U.S. elections in their native language.  There could be all kinds of things going on in people's lives.  You don't know. 

Sometimes I think people in group (6) are the healthy people, and those of us who think about politics all the time... well, there's something wrong with us.  (As Eliezer Yudkowsky used to say, "Politics is the mind-killer.")

Anyway, I suspect groups (2) through (6) exist and are quite large and groups like them will continue to exist for many more election cycles.  In addition, there is probably another group this particular time, namely:

7) People who don't want to vote for anyone who might be guilty of rape or sexual assault.  Since that may be true of both candidates this election cycle, this puts such voters in a difficult position.

And undoubtedly we could come up with more groups if we thought about this harder.

The problem we have with "swing voters" is that, when we picture the "swing voter," we imagine group (1), and that probably seems silly, or inconceivable.  But think more about the other groups.  You may not agree with the people in these groups, but you are probably aware that they exist.  Their beliefs are imaginable, and, if you do a little of the work of empathy, probably mostly understandable.  You can probably at least put yourself in the mindset of people in groups (2), (3), (4), and (7). You probably know people, personally, in all of these groups (except group (1)). 

This was Jon Favreau's mistake.  When he spoke to focus groups, they kept saying things that put them in camps (2) through (7), but no matter how much he listened, all he ever heard was (1).

Another complication here is that a person may appear to a pollster to be in one group, when they're actually in another group.  Let me be clear here: I'm not referring to the "shy Trump voter" problem.  As I've already indicated, I don't think this is as much of an issue as many people do.  But there are, for instance, probably many people who would respond to a pollster's questions as if they were part of group (4), and who may even think of themselves as belonging to group (4), but, who, at the end of the day, reveal themselves not through their statements but through their actual behavior to belong to groups either (2) or (3).

There are many questions to be asked here, and much real research to be done. I, for one, particularly want to know: 

• How large is each group?  I particularly want to know: which is larger - group (2) or group (3)?  My guess is that groups (2), (3), and (4) are much, much larger than all of the other groups combined, and group (1) is virtually non-existent, but I could be wrong.
  
  
• For each group, how likely are the people in it to vote?  My guess is that groups (4) and (6) often don't end up voting, whereas (2) and (3) mostly do.  I'm not sure about (5) or (7).  On the other hand, maybe we can break down things even further, especially in the case of (4).  They may say that they have no opinion about which is better of two candidates because they're disgusted with the entire system... but when election day finally arrives, some may stay home, but some will, grudgingly, make up their minds.
  
  
• How persuadable are the people in each group?  This is really the key question.  My guess is that, (2) and (3) (and possibly (4)) are not very persuadable - at least, it's extremely unlikely that you could persuade a person in group (2) to vote for Trump, or a person in group (3) to vote for Biden.  You might be able to persuade them to vote for a third party, or not to vote at all.  And since I think group (4) is unlikely to vote, it's probably a waste of your time to try to persuade them.  But the key idea here is that "undecided" and "persuadable" are by no means synonyms.
  
  
• And... if they are persuadable, how can they be persuaded?  This is an important question, too, obviously, and I think the answer will depend very much on which group they're in.  Some of the people who are further to the left than the Democratic candidate can probably be persuaded to vote for the Democratic candidate, and some of the people who are further to the right than the Republican can be persuaded to vote for the Republican.  In my opinion, that's really where the most energy should be spent.  

And those who think that the government is controlled by space aliens can be persuaded by convincing them that the aliens favor your candidate.... or that they don't.  Whatever.

So what do we know about these people?  Not much.  Unfortunately, statisticians just squish these people, who are all very different from each other, into the meaningless blob of a category, "swing voter" or "undecided voter".  They need to do more careful, precise work here.   There's a lot of research to be done.  Our democracy depends upon it.