R for Political Data Science Week 8: Four Parties in America? Probably Not Anytime Soon

This is part of a series of short posts about politics that seeks to show how we use data science to learn more about the real world. Follow along here.


The time has come: I have become one of those people that reads the New York Times Opinion page regularly just to find hot takes to rebut on the internet. This week, Thomas L. Friedman has an Op-Ed asking “Is America Becoming a Four-Party State?” that promotes some … dubious… understandings about American politics. I lay out a quick reading of his article for you here and provide an abbreviated rebuttal.

Friedman argues: Because of ideological diversity on the left, and an ongoing split on the right (conservativism vs Donald Trump), America has the capacity for four political parties.
However, these divides are exaggerated in (a) magnitude and (b) frequency among the public. Plucking a few left-leaning/establishment Democrats and Trumpist/conservative Republicans is not the proper measurement tool. In polls, the divide is muddier — and ideology matters less.
Even if Friedman was correct there were a large block of Democratic voters that are as liberal economically as they are socially — like Alexandria Ocasio-Cortez, his favorite example of the new Democratic party — they wouldn’t abandon the party; loyalty to the label is just too strong compared to the importance of ideology in voting behavior.
Finally, EVEN IF a substantial number of uber-liberal Democrats and anti-Trump Republicans peeled off to form a third and fourth political party, the US electoral system is not designed to support more than two parties. The combination of a first-past-the-post electoral system and Electoral College means that a vote for someone other than a major candidate is just throwing your vote away.
I’m going to crunch some numbers to illustrate my point. Let’s look at ideology and partisan loyalty, the latter measured by whether or not someone feels represented by their political party.

Looking at data from the 2018 VOTER Survey from the Democracy Fund’s Voter Study group, the evidence is clearly against Friedman. The vast majority of most voters, regardless of party or ideology, feel well-represented by their party. See this graph, where each point is a voter placed along the left-right ideological spectrums for economic and social/elite attitudes and colored by whether or not they feel represented (green) or not (brown) by their party:


It’s just not the case, as some would argue, that moderate Democratic voters feel “left behind” by a Democratic party that is leaning more to the left year after year, or that those liberal Democrats might start their own party. The entire argument strikes me as the same misundersttandings of the Tea Party come back again, but on the left. And that’s if we’re equating the size of the two movements, which of course is not true.

Besides, we know from a plethora of political science research that a voters’ issue-based ideological proximity to a party — whether a voter who supports raising the minimum wage can find a candidate who supports raising the minimum wage — is not the end-all be-all for determining which party they will vote for. In fact, it’s not even close. Aside from the fact that your party affiliation can predict about 95% of your voting behavior, demographic characteristics such as race, age, income, and education as well as attitudes like racism, sexism, feelings toward your economic security and so-called “populism” are all predictive of vote choice.

The assertion that we could have four political parties in next year’s presidential election simply falls apart when stacked up against the evidence.

There’s more work to be done here, but I think this graph illustrates the point clearly enough to set aside more discussion in other channels (IE: my real job?).

If you’re here to learn R, be sure to check out the code on GitHub.

R for Political Data Science Week 12: Do Voters Still Care About The Economy?
R for Political Data Science Week 11: Is Beto the Media Sweetheart?
R for Political Data Science Week 10: What If Each State Allocated Their Electoral College Votes Proportionally?