# How does turnout in US elections compare across states?

It’s well known that many US states vote in a very predictable way in national elections (e.g. Arkansas has a Partisan Voting Index (PVI) of R+15, two Republican senators, an all-Republican House delegation and a Republican governor, while Hawaii has a PVI of D+18, two Democratic senators, an all-Democratic House and a Democratic governor) while others are considered swing states that might vote either way (e.g. Pennsylvania with an even PVI, one senator of each party and an evenly split House delegation).

I’m curious how these differences affect the turnout in nationwide elections. Naively, one might expect swing states to consistently have a higher turnout as every vote may matter while in one-sided states many voters may think that their vote does not matter. However, I believe there are usually many different elections happening at the same time on Election Day so this simplistic view may not hold as other (local or district) elections may draw voters in.

So how does turnout in national (presidential, Senate and maybe House) elections compare across states? Does turnout correlate with any political indicators?

• Arkansas is not quite so predicatble, it had a democrat governor as recently as 2015, and does anyone remember a guy called Bill Clinton? Commented Aug 29, 2020 at 17:37
• The last question about "any political indicators" is quite vague and open-ended. Are you asking about the correlation between turnout and the PVI or are you asking about something else entirely? Commented Aug 29, 2020 at 18:39
• @BrianZ The working hypothesis was that turnout correlates with contestedness but I tried keeping that sentence vague and open to allow for all sorts of correlations that the data might contain to be added to an answer.
– Jan
Commented Aug 31, 2020 at 4:39

Looking at the correlation between Cook PVI & average difference from the national turnout in national elections over the 2012-2018 period, we can plot the graph below, which clearly shows the trend identified in the question. On a very rough basis, the trend line shows that every PVI point corresponds to a reduction in turnout of 0.65 pp against the national average turnout. The Pearson correlation coefficient is around -0.46, indicating a strong negative correlation between the two. Red points indicate states with Republican PVIs, blue points correspond to states with Democratic PVIs.

If we exclude the outlier of DC, with its D+43 PVI, and plot the points on scale where Democratic PVIs are negative, we can fit a polynomial to the data which leads us to similar conclusions.

I'm sure someone will come along with a more sophisticated analysis, but here's a quick an dirty one for now, and it does suggest a correlation between high turnout and low partisanship.

Looking at the data from the US Elections Project I picked out the list of states which had both over 64% turnout in 2012 and over 45% turnout in 2014 (measured as ballots for highest office as a share of eligible population). For those states, I took the Cook index from Wikipedia (which are for 2012-2016).

High turnout states:

• Iowa R+3
• Maine D+3
• Minnesota D+1
• New Hampshire EVEN
• Wisconsin EVEN

Now compares that with states which had below 56% in 2012 and below 32% in 2014:

Low turnout states:

• California D+12

• Indiana R+9

• Tennessee R+14

• New York D+12

• California D+12

• Indiana R+9

• New York D+12

• Oklahoma R+20

• Tennessee R+14

• Texas R+8

• Utah R+20