The Atlantic picked up on an interesting piece of research from a demographer called Conor Sen from Atlanta:
The whole article is well worth a read. Something that sticks out however, is this claim:
But you may also be struck by the shape of that trend line (Sen is quick to note, by the way, that he’s not a statistician). It roughly suggests a political tipping point somewhere around a population density of about 800-1,000 people per square mile.
Justin Esarey, Assistant Professor of Political Science at Rice University, has picked up on it and noticed something quite interesting.
Huh. Well, that did not look like a very good fit to me. So, I reconstructed the data set using the Wikipedia-sourced PVI data and the Census-sourced population density data that Conor talked about. I then ran an analysis of this data replicating Conor’s log-fitted model, plus a loess nonparametric fit line and a simple linear model. Here’s what these three models look like when plotted against one another:
What Justin is showing, is that by simply employing a different type of trend line you can paint a very different picture.
The interesting part (for me!) is how this would actually apply to political campaigns. For example, Dave Troy came to this conclusion using the original log fit line:
at about 800 people per square mile, people switch from voting primarily Republican to voting primarily Democratic
If you were to base a campaign on that principal, then depending on your voting system, it could be a fair assumption for a Democratic campaign to ignore districts with a population density of less than 800 people per square mile, and likewise, that very dense districts are only marginally more Democrat-leaning than moderately dense districts.
The two other trend lines tell a story. With them you get a far simpler, and in my view, far more logical story. Whereby there is no “tipping point” where a low population density district area becomes “worthless” to Democrats, and where a district continues to be stronger Democratic the higher the density.
All goes to show how important it is to get statistics right.
Please note: I am in now way claiming to be a professional statistician at all, so please excuse any errors on my part too!