Of course, I have no idea how anyone can win any election, but that’s what this post is about anyway, and no one ever got clicks by admitting they aren’t sure. This is just a little election nervous energy, dedicated to David Brooks, who says “data driven” campaigns have been a fiasco. (Follow this series at the elections tag.)
With all this talk of political realignment, and the impending fracturing of the Republican Party, I was wondering how the parties might shake out if something serious really did happen. I struck me that the GOP establishment is really upset right now because the beast they have been cultivating (the racist, xenophobic, sexist, anti-gay, anti-science beast) is suddenly turning on them in way that threatens the things that really matter: keeping taxes and regulation from impeding their accumulation of wealth. As if what they really want is their mass of gullible voters back from Trump.
Anyway, I looked at the American National Election Studies 2016 pilot data, which asked 1,200 people a lot of attitude questions, as well as how they feel about the candidates. My question was, how are left and right divided now, and what might happen if things broke differently.
The results are all in the figures below. To make these I used Stata’s cluster kmeans command, which breaks a sample into the number of clusters you specify according to the distribution of means on the variables you list. Using math. This is nice for politics because it forces each person to choose a cluster, just like voting for a party. I used the feelings and issues you see in the top part of the figure to create the clusters, and then checked to see how the members of each cluster feel about the candidates. I ignored voter demographics, just taking into account their opinions. I used Excel’s conditional formatting to color-code from green to red (but check the coding because left and right don’t run the same way for each question — note if the item is “oppose” on a scale of 1-7, then 3 or less is “favor”).
The first figure is a two-cluster scheme. Every potential voter has to pick a party. In the two-party scheme a few major differences stand out. The left feels much better about scientists, gays and lesbians, and feminists. They are more interested in action on climate change, paid leave, equal pay for women, and big government generally. They favor affirmative action at universities and raising the minimum wage. They think legal immigration is good. Issues that don’t divide left and right so much are feelings about the police, free trade, and crime. The death penalty is not huge. The left’s favorite person is Sanders, but they like Clinton, too. They hate Trump. Here’s the first figure (click to enlarge):
The left here is only 45% of potential voters, so come November they need to raid the right somewhere – but where? To answer that I specified four clusters instead of two. This gives the left and right two center categories to fight over. Now the two left parties together sum to 45%, so for the left to win, I figure they need to find issues they can use to pull people away from the right, so I’m especially looking for big differences between the right and center-right. The good news for the left here is that the right is only 22%, so there is a lot of potential for poaching the center-right. In this figure, I highlighted a few key left-right divisions arbitrarily. Here’s the figure, with my comment below (click to enlarge):
Fortunately for the left, the right is blinded by their hatred of gays and lesbians and feminists. The center-right is much more tolerant, especially of gays and lesbians. So that seems like a good wedge. I say “tolerant” because the center-left isn’t really crazy about gays and lesbians or feminists, so going all the way to awesome in that area might be a turnoff to them.
The other good ones for the left are equal pay for women and legal immigration. The right hates them but everyone else is positive. Science is also good for the left, as the right doesn’t like scientists much, and the center-right is very into vaccination.
On the other hand, the center-right is very pro-police, so that may not be a winner for the left. The anti-police right isn’t coming all the way over, so it doesn’t help that they’re not pro-police. Also, affirmative action and the minimum wage aren’t good for pulling out the center-right. In general, the Chamber of Commerce type issues — climate change action, free trade, and big government, don’t help the left split the right much.
One caution is that the center left doesn’t like Clinton much — no better than they like Rubio, in fact. They are more positive about Sanders.
Obviously, how to work this into a campaign — with all the other issues that entails — is way beyond me. That’s why this is free.