Do Libertarian Votes Reduce Imprisonment Rates? Incarceration in Red States and Blue States

Do Republican states incarcerate more people than Democratic states? Or vice-versa?

In DataSplash, we explore whether there is a relationship between state-level support for the Democratic Party (as compared to the Republican Party) and the number of people incarcerated in the state.

To get more specific, voting for a Republican presidential candidate, as opposed to Democratic presidential candidate, is in part due to culture. Thus voting patterns can be thought of as an indicator of culture. To the point: Did the red states throw the book at prisoners any harder than the blue states?

Using the following data, we plot state prison admission in 2012 — on the vertical axis — against state percentage 2012 vote for Obama — on the horizontal axis.

Using the drag and drop feature, you can import the data yourself into DataSplash and see if what I’m doing is right, whether it makes sense, and whether I’m overlooking something important! — I welcome any feedback.

DataSplash by default puts the best linear fit (aka regression line) through these points, once you’ve loaded, or dragged and dropped, the data set into DataSplash. If you do so, you should see the same as the screenshot above.

Visual inspection shows that this linear fit line is basically flat. While visual inspection of the red line suggests that there is a slight upward slope, indicating more voting for Obama is associated with more incarceration. The result is confirmed by the numbers in the Regression Results tab below the scatter plot. However, the Regression Results tab also tells us that, statistically speaking, the correlation between the Obama vote share and incarceration rate is zero.

Texas is a large outlier. What happens if we remove that data point? The pattern changes noticeably, i.e. the new linear best fit gets steeper:

The dashed line is the previously estimated line, when we had included Texas. The new line shows a larger positive association, but it is still insignificant, as shown in the Regression Results tab.

Note that the Regression Results tab also displays the previous estimate that had included Texas, and the current estimate.

Is it possible that party is such a weak indicator of differences in culture? Perhaps we find no difference between red and blue states, because the two main parties have similar positions on incarceration!

To figure out whether political views, besides being a Democrat or a Republican, matter for the locking up people behind bars, we explore the vote share obtained by Gary Johnson. Johnson was the presidential candidate for the Libertarian party in 2012 and who also happens to be the candidate for the Libertarian party in 2016. Johnson may never win, but he often claims his goal is to send a message. Perhaps we will hear a piece of his message when we relate his state vote share to state prison population.

When we load the data for all 50 states, and remove our outlier, Texas, we discover a significant relationship, as shown in the screenshot above. The trend line indicates that a 1% increase in the vote for Gary Johnson in 2012 was associated with a decrease in prison admission by roughly 4,000 people. Although, we find no significant correlation when we include the Texas outlier- something I do not show here.

What does the result that the Johnson vote share is associated with a decrease in incarceration mean?

Perhaps it is evidence for the optimistic idea that third party candidates can send a message and can influence policy, even with a small vote share.

More likely, this is evidence that voters who cast a ballot for a libertarian candidate are not fond of the large-scale incarceration that we’ve seen over the past twenty or so years. And these voters are making their voices heard, and affecting political choices in their states.

I am sure, there are many other stories explaining this downward sloping read line, and I’d love to hear your stories and theories!