Wednesday, December 21, 2016

Donald Trump Won States With Lower Voter Turnout

A common refrain in the aftermath of most elections goes something to the effect of  "the People have spoken." That's how democracies work, right? Whoever earns the mandate of the people gets the position. Simple. But considering that Hillary Clinton is leading by over 2 million votes in the total vote count, we know that the concept of "the people" is a bit wonky in US elections. It's not, as has been bludgeoned over our heads in the last couple of weeks (sometimes by ourselves), a popular vote system. The president is decided based upon the electoral college vote and that institution, for better or worse, has enabled Donald Trump to be the President Elect. At the very least, this election will serve as a fantastic example for civics classes for years to come.

One of the most common arguments for the electoral college is that it preserves the diverse voices and issues of the states. California is not like Hawaii is not like Kansas is not like Iowa-- which is kind of like Indiana considering that in both states there's nothing but corn.

Forever. (Source: Wikipedia)

The point being that these states have diverse issues that need to be addressed when dealing with an entire nation-- especially one as geographically and culturally diverse as ours. It's a legitimate concern and I discuss it in my latest video on the Professor Politics channel. Ideally, then, the system is understood as "the people of the various states which make up our Republic have individually spoken and their results are collated to ideally maintain the tenuous balance between the state and national power." But his is obviously a bit of a mouthful so "the People have spoken" usually gets the point across.

The concept of the people having spoken is predicated on one teensy, little, condition, though: The people actually, you know, speaking. It's kind of hard to say that anyone has the mandate of the people when the largest plurality of the vote went to "not going out to vote." So since this is supposed to be the people of the states speaking, I wanted to see if there were any differences between the voter turnout in states that Trump won or that Clinton won. And since you presumably read the title of the article before clicking on it, I feel safe in spoiling the fact that, yes, there is a difference between the two. Trump did better when fewer people voted.

Now, out of context, that sounds like a bit of a no brainer. A number of people have been rallying around the idea that Democrats do better in larger states. This is why people decrying the popular vote argue that California and New York would just steamroll over everyone. They are presumably unaware that Trump won 7 of the 10 largest states in the country but, hey, that's none of my business.

But here's the kicker: This is by proportion of the states Voter Eligible Population (VEP). So large states, small states, blue states, red states-- all are judged on the same criteria. The proportion of individuals who did vote compared to those who can. I pulled the VEP numbers for 2016 from the United States Elections Project and the turnout numbers from the same source (although, if I'm not mistaking, they ultimately originated from the states' Secretary of State's offices). I then split the states into two categories: Trump V. Clinton. The average turnout rate for a state that Trump won was 59.7%. The average for Clinton on the other hand, was 63.5%.  The p-value for this difference was .056-- so we can be fairly confident in this difference[1].

Here's a cartogram of the United States that's generated by the proportion of the state's VEP that turned out to vote[2].

Here it is again, except this time colored by whether or not Trump won the state.

You can see that those that have been shrunk and contorted the most tend to be red. Those that more closely retain their shape tend to be blue. It's definitely a tendency thing-- there are clear exceptions. Florida looks more bulbous than usual and Vermont definitely shrank a bit. However, the general pattern is clear. Those states that went Trump, by and large, are, well, not large.

Thank you, thank you. I'll be here all week.

A word of caution: Some may be inclined to connect this map with things like North Carolina, where the Supreme Court ruled that Republican lawmakers were effectively engaged in vote suppression or to other patterns of the sort. This map does not make any claim to voter suppression or vote tampering; there is not enough here for those suppositions to be reasonably made let alone for a conclusion to be drawn. It could very well be that only half of the eligible people in Texas voted because it's a "safe" state ("Why vote? It's going to go red anyways."). People may simply have just stayed home. The point I'm trying to draw across here (besides not-so-subtly showing off my limited GIS skills) is that it's even hard for people to say that "the people have spoken" when, in fact, the states that did more of the speaking tended to go for the person who lost. We also don't know the characteristics of who stayed home. Were they Democrats? Republicans? Neither? All we can show here is that the states that Trump won exhibited this pattern.

These questions raised by the analysis I imagine that's where it gets interesting.

This post originally appeared on my Medium account. Check it out here.

1: For the more technically minded, I used a two-tailed t-test accounting for unequal variances. Sample size of Trump's wins was 31 while Clinton's was 20. I feel confident in utilizing an alpha of 0.1 due to the sample size but, honestly, it's a knife-edge result as is.
2: Again, for the technically minded, this was generated using the Cartogram package in QGIS. I ran into a bit of trouble with it that I'll explore in another post (how else am I going to keep up with my production schedule if I don't at least have a couple of easy weeks?).