How To Try To Think Clearly About The Voters Who “Decide” Elections.

(Hint: Which Justices “decide” Supreme Court cases?)

There’s a lot of talk these days about which voters will “decide” the Presidential election. Unfortunately, a lot of it confuses me. There are over a hundred million Presidential voters in the United States and it’s very hard to think about numbers that large.

When all nine Supreme Court seats are filled, there is also a lot of talk about which Justices will “decide” various cases. That’s much less confusing. Barring any recusals, there are only nine Justices on a filled Supreme Court. It is much easier to think about nine people than to think about a hundred million people.

Consider, for example, the Supreme Court case “NFIB vs. Sebelius”, where the Affordable Care Act (ACA) was mostly upheld. I am not a Supreme Court expert but I think these things are widely understood:

1. Prior to the ruling, as far as I know everyone understood that what mattered was which side would get at least five votes out of nine.
2. Everyone knew that the four Democratic appointees — Sonia Sotomayor, Elena Kagan, Ruth Bader Ginsburg, and Stephen Breyer — were all but certain to vote to uphold the ACA.
3. Everyone knew that the three most conservative Republican appointees — Samuel Alito, Antonin Scalia, and Clarence Thomas — were all but certain to vote to overturn it.
4. So everyone knew that the case would effectively be decided by the two less conservative appointees, John Roberts and Anthony Kennedy.

In other words, everyone understood that the case would be decided by the Justices at the median support level, because they were the Justices who would get their side over 50%.

When Roberts eventually voted with the four Democrats, everyone acknowledged him as basically responsible for the case’s outcome, even though in some sense he was just one vote out of five and a very atypical member of the winning majority:

(Actually I see on Wikipedia that Kennedy was a leading opponent of Roberts’ decision behind the scenes, but he was at least “thought to be the administration’s best hope to provide a fifth vote to uphold the law” beforehand. Again, this is just an example for illustrative purposes; even so I’d be surprised if any of Alito, Scalia, or Thomas were a likelier vote to uphold than Kennedy.)

There are only nine Justices so it’s often easy to see who the median Justice is. As far as I know, no one tried to figure out the case by dividing the Justices into demographic groups and then looked at which demographic groups “uphold” won and called them “the ACA coalition”. The male Justices happened to vote 2–4 to overturn while the female Justices voted 3–0 to uphold, but no one made a table like the one below, or wrote thinkpieces about how Supreme Court Democrats don’t need male Justices anymore because “the winning margin came from female Justices”:

But as I said, it’s easier to count to five than to count to sixty-four million. People really do look at tables like this, from the 2012 exit polls, and write thinkpieces about “the Obama coalition” of demographic subgroups Obama won and how Presidential Democrats don’t need white voters anymore:

While this is useful information (or it would be if it were correct, anyway — it’s likely that there were more and more Democratic white voters than the exit poll indicated), it doesn’t tell us what we should want to know any more than the previous table did.

One thing we should want to know is: Who were the median voters? Which voters, exactly, got Obama over 50%, either nationally or in the various swing states?

What if instead of a table of subgroups, the default way of visualizing an electorate was more like this (which, yes, uses the Josh Katz/Nate Cohn NYT estimates instead of the exit polls):

All three demographic groups are split into the portions that voted for Obama and Romney and sized proportional to their share of the whole (two-party) electorate.

Maybe visuals like this would make it clearer that the voters who got Obama over 50% weren’t necessarily members of groups that supported Obama. In fact, I feel like “the Obama coalition” is often described as “minorities and educated whites”, but Obama would have only gotten to 33%-40% without the non-college whites who supported him, depending on if they stayed home or switched to Romney. (Maybe we wouldn’t have spent years talking about the electoral irrelevance of non-college whites!) Even using the exit polls, Obama only would have gotten to 39%-45% without non-college whites.

More specifically, millions and millions of white men without bachelor’s degrees voted for Barack Obama in 2012, even though non-college white men were strongly Republican overall. According to Josh Katz and Nate Cohn’s estimates in that link, Obama received 9 million votes from non-college white men compared only 7 million votes from, say, Hispanics. I’m not saying it’s true, but it’s possible that among those 9 million votes were the ones that got Obama over 50%. His national margin was only 5 million votes, so he certainly wouldn’t have gotten over 50% without any of them.

Some of those Obama-voting non-college white men might have been the John Robertses or Anthony Kennedys of the 2012 election, if you will: effectively deciding the outcome even though similar voters mostly voted for the losing side, and with no guarantee they would vote the same way next time. Even with that possibility, people rarely if ever include them in “the Obama coalition”, because a lot of other non-college white men voted for Mitt Romney, and because they trended Republican as a group etc.

It’s not a perfect comparison. Perhaps the non-college white men who did vote for Obama were strong liberals, and his marginal supporters were elsewhere. Although just because non-college white men as a group trended Republican does not necessarily mean that the ones who still voted Democratic were assured.

And in actual elections, people look at turnout, not just support. Maybe the voters who got Obama over 50% weren’t non-college white men who almost voted for Romney but young people or minorities who almost stayed home (or almost voted for Romney).

Of course median voters change from election to election. The median voters of the 2016 election might turn out to be traditionally Republican suburbanites who almost voted for the GOP but hate Donald Trump — or they might turn out to be some of those non-college whites who voted for Obama but hate Hillary Clinton.

I don’t know, and I don’t know how to predict the election. But I do know that a lot of writing and thinking about it would be a lot clearer if we used the same approach we use with smaller “electorates” like the Supreme Court: Trying to figure out who the median voters are and what they’re going to do.

This applies to other elections as well. After months of everyone writing fervidly about nearly every aspect of the Clinton/Sanders primary, I know that young people were strongly for Sanders and African-Americans were strongly for Clinton, but I can’t even think of anyone who even asked who the median voters were, let alone tried to answer. Jamelle Bouie, for example, said “HRC won b/c dominance with black voters” without specifying why he’s sure black voters were, in the aggregate, more like the John Roberts of the primary than the Ruth Bader Ginsburg of the primary.

( The closest I can think of is the writer Matt Bruenig, who mused at some point that Sanders didn’t necessarily lose because his voters were “too white” because he might have won with more white voters. Not quite the same thing, but in the right direction. Also, I’ve wanted to write this piece for a while, but was particularly motivated by a Twitter exchange yesterday with Josh Barro and Nick Riccardi, who made some similar points.)