Author Archives: Boris Shor

Polarization in the West, 1996-2017

I previously posted new visualizations of political polarization in Southern American state legislatures in preparation for a conference honoring the work of legendary political scientist Keith T. Poole. Here’s those same plots for Western state legislatures from 1996-2017 (marked in 4 year intervals plus 2017 preliminary data). As always, polarization is measured as the distance between the Republican and Democratic party medians, averaged across the two chambers. Lines are colored according to which party is polarizing faster.

States in the West are the most polarized, and the fastest polarizing, according to our data. Colorado, California, and Arizona dominate the race to the poles, though other states too like Utah are quickly charging up the rankings.  Colorado only in the past few years has overtaken California for the title of the most polarized state legislature in the country. Hawaii is clearly an outlier, with a liberal Democratic party paired with a liberal Republican party (in the House).

In terms of asymmetric polarization, here too, with the exception of Colorado, Democrats and not Republicans have been polarizing in more states.


Here are the positions of Republican party medians in the West. They’ve polarized quickest in Colorado.


The Democratic slopegraph is more dramatic. Dramatic falls are visible in almost every state except Alaska and Idaho.


Polarization in the South, 1996-2017

I’m working on new visualizations of political polarization in American state legislatures in preparation for a conference honoring the work of legendary political scientist Keith T. Poole. Here’s a slopegraph (inspired by Edward Tufte and powered by Thomas Leeper’s awesome R package) of polarization in Southern state legislatures from 1996-2017 (marked in 4 year intervals plus 2017 preliminary data). This was a period of intense transformation in the South, when the almost-fully Democratic-controlled legislatures transformed into Republican-dominated ones in just two decades. Polarization is measured as the distance between the Republican and Democratic party medians, averaged across the two chambers.

In every single state, the parties have been polarizing. And would you just look at Texas? It was the most polarized in 1996, and it’s only increased its lead in 2017.

The line colors represent whether Republicans or Democrats polarized faster in that state. And, in contrast to Congress, Democrats are polarizing faster more frequently than Republicans in the South, including Texas.


Here are the plots for the Republican and Democratic party medians in the South over this time period to give you a sense of how the parties changed. First the Democrats:


Now the Republicans:


Updated Polarization Plot for 2015-2016

We plan to update our publicly accessible data to include the results of the 2014 election.

In the meantime, here is how the states stack up in terms of polarization. One big change has been the vast increase in polarization in Colorado, which nearly matches California’s polarization. Other rapidly polarizing states include Arizona, Texas, and Missouri.


June 2015 Update to Shor-McCarty State Legislatures Data

Today I posted the individual, legislator-level ideology data as well as the aggregate state and chamber-level data for our state legislature ideology data. The data is free for public download and use. Please cite our data and paper if you use this in your academic work, and/or link back to us in other work.

We have extended the data collection through 2014, and we now have 20,783 unique state legislators in our data, as well as 1,816 chamber-years of aggregate data.

As with previous iterations, we have continued to spend a lot of time and effort cleaning up the data; there should be far fewer noise-related issues. But please let us know if you see particular problems.

You can find the individual data here and the state level data here.

Bugs should be reported to us.

The roll call data collection in this update has been supported financially by the Laura and John Arnold Foundation.

The New New York State Assembly Speaker is Pretty Unusual

Sheldon Silver, New York State’s iconic Assembly Speaker, was forced to resign after being charged with serious federal corruption charges. Bronx Democrat Carl Heastie is going to ascend to the position sometime on Tuesday, after his last competitor, Queens Assemblywoman Catherine Nolan, dropped out of the race to succeed Silver.

I tweeted on Saturday that I was surprised by this turn of events. This was after another competitor, Democratic Assembly Majority Leader Joseph Morelle (apparently the favored candidate of Governor Cuomo), had dropped out of the race.

I was surprised because Carl Heastie is fairly extreme ideologically. Yet almost all the attention in the discussion about the speaker’s succession has been about race, potential ethical issues, and a real or imagined power play between Governor Cuomo and Mayor de Blasio.

Our state legislative ideology data set provides me with the data for this post. The current release of the data has data into 2013, but we are currently working on extending that data set into 2015. I don’t have the newest members serving in 2015, but they are relatively few as the chamber isn’t very competitive. So I’m just using the scores and composition of the chamber as of 2014.

Sheldon Silver is in the most liberal fifth of his party in the Assembly. Nolan has basically the same ideological profile as Silver. In contrast, Morelle is a little to the right of the median Democrat. Carl Heastie, on the other hand, is the fifth most liberal Democrat (only Weinstein, Rosenthal, Glick, and Robinson to his left) in the Assembly, out of a total of 106. And it’s not like Silver, Nolan, or even Morelle are hardline conservatives. The New York State Democratic party is amongst the most liberal in the country, alongside California and Vermont’s Democrats.

This is a puzzle because I’d expect New York Democrats to pick someone closer to their ideological center. This should be due, crudely speaking, to the median voter theorem. A simple majority-rule election should have the property of privileging centrist candidates. Just look at the 113th Congress US House leadership team of John Boehner, Eric Cantor and Kevin McCarty, who were just about at the center of their party in the House.

A recent paper by Stephen Jessee and Neil Malhotra shows empirical evidence for this notion: party leaders in Congress tend to be not very far from the center, especially in comparison to their competitors for the offices (Democrats do shade somewhat to the left and Republicans to the right of their party medians). In this case, however, the winning candidate is especially far relative to not only the party as a whole, but in comparison to his announced competitors.

What’s going on here? Are states that different from Congress? Or is something peculiar about New York in particular? I’m not sure, but events like this get me thinking about new research projects …


Updated (July 2014) State Legislator Ideology Scores Posted

Yesterday I posted the aggregate, state-level ideology and polarization data for our state legislature ideology data. That is actually based on legislator-level data. Today, we are posting and making public and free that individual level data.

We have extended the data collection through 2013, and we now have 20,198 unique state legislators in our data, or about 11% more than the previous update from May 2013.

As I mentioned yesterday, we have spent a lot of time and effort cleaning up the data; there should be far fewer noise-related issues. But please let us know if you see particular problems.

New this year is a district-by-year indicator, as these legislators do change districts over time.

You can find the data here.

Bugs should be reported to us.


State Legislative Ideology and Polarization in 2013-14

Today we posted the July 2014 update to our state legislative aggregate ideology data. One of the major features to this data release has been the incorporation of much more current roll call (1993-2013) and candidate survey (1996-2014) data. All in all, we have added about 25% more chamber-years of data.

Here’s a look at the average legislative ideology  nearly all 50 states that we have data on in 2013-2014 (the data is from 2013 but nearly all states have elections in even-numbered years, hence the data apply to 2014 as well). Higher numbers are more conservative, lower numbers are more liberal. In 2013, Oklahoma had the most conservative state legislature, followed by Arkansas, Tennessee, and Idaho. If Congress were a state, 18 states would be more conservative than it!

New York was the most liberal state legislature in the country, followed by Connecticut, California, Vermont, and Hawaii.


And the equivalent plot for polarization in 2013.

California, once again, leads the country in the ideological divide between the Republican and Democratic parties. Colorado and Arizona are close behind, gaining ground over time.

At the other end, Rhode Island, Arkansas, and Louisiana represent the least polarized legislatures. Note that Nebraska is not in this group. This is despite the fact that the Unicameral is officially nonpartisan in its organization and elections. Nebraska’s hidden parties are polarizing quite quickly.