Image from: https://web.stanford.edu/~neilm/
When last we chatted, I talked of how congressional elections have shifted to reflect national politics more than local influences. Importantly, I noted that this may be less the result of ideological “polarization” — the process by which the beliefs of partisans move left/right — and more the result of partisan “sorting.”
But polarization does exist, and it is real. This is especially true of elites in government: congresspeople in particular.
But there is a wealth of research on the subject, and there’s not much new on which to write a blog post. Instead of summarizing all of the research here, I aim mostly to use these trends in a geeky way, using a cool new graphics package for the R statistical software called ggjoy. If you want this short post in tweet form, here you go.
First, some of the always-relevant work on polarization. Just for fun:
- Keith T. Poole and Howard Rosenthal, “D-Nominate after 10 Years: A Comparative Update to Congress: A Political-Economic History of Roll-Call Voting”
- Pew Research Center, “Political Polarization in the American Public”
- James E. Campbell, “Polarized: Making Sense of a Divided America”
- More, by way of my reading list for US political ideology.
Now comes the fun part. What exactly is ggjoy? Ggjoy is a component package for the common graphics library ggplot. Ggplot is responsible for 99.9% of all the graphics I make on this blog. Ggjoy allows us to plot distributions of data over intervals of time. We could, for example, plot the average temperature for each month last year in Lincoln, Nebraska. That looks like this:
The graph above tells us that Julys in Lincoln are relatively predictable, usually between 65 and 85 degrees Fahrenheit. December, on the other hand, has some pretty extremely cold weather rather frequently.
That’s all good and fun, but I don’t run this blog for the weather. What can ggjoy help us learn about political polarization?
Ian McDonald, a visiting professor at Lewis & Clark College in Portland, Oregon had the incredible idea to graph the distribution of House members’ ideology over time. Using the distribution of a mathematical representation of ideology called DW-NOMINATE, McDonald showed how Republicans have moved right on mostly-economic policies over time, while Democrats have generally stayed ideologically the same.
I have added to his graphic (which appears on the left below) with a second mathematical representation of ideology in DW-NOMINATE – a so-called second “dimension” – to show how Democrats have changed in ways McDonald’s graphic does not show. This second dimension is often referred to as a “social” dimension, mainly encompassing the shift in views towards race that started to flip the parties in the last 1960s.
There is a lot to be seen in the ideological movement of the parties since the 1960s, much of which the links above cover in very great detail. As you can see, ggjoy is a good package for efficiently summarizing distributions of data over time.
I think my next ggjoy plot will be of congressional vote shares over time. We know that districts are moving farther left and right than they ever have before, and the time-series distributions will undoubtedly yield itself well to demonstrating that.
In the meantime, what are you interested in? Send me a tweet and I’ll get right back to you! Until next time,
Thanks for reading today everyone. Tune in to my twitter for updates and make sure you sign up for my newsletter to get notifications of recent posts.
- R for Political Data Science Week 6: Just How Liberal Are the 2020 Democratic Candidates?
- R for Political Data Science Week 5: The Ideological Diversity of the American Electorate
- R for Political Data Science Week 4: What Happens To Our Algorithms When Socialists Vote in Congress?