Skip to main content

2013 Was the Year of the Dataset

Large collections of information can be used by more than just the NSA. They're being assembled by scholars and researchers hoping to find a better way to understand American politics.
  • Author:
  • Updated:
(Photo: S.John/Shutterstock)

(Photo: S.John/Shutterstock)

This past year may end up being remembered as the year that we discovered that the NSA and telecommunications companies were gathering enormous troves of personal data on us and using it for ... well, we don't quite know what yet. But on the less nefarious side, 2013 was also the year that saw the public release of several groundbreaking datasets that allow scholars and researchers to better study American politics. They are enabling those interested in understanding politics to move away from conjecture, hunches, and narratives, and toward actual empirical measurement. Below is a quick discussion of four of these new datasets:

The Shor/McCarty State Legislator Ideal Point Data
We've had pretty rigorous methods for converting the roll call votes of members of Congress into "ideal points" (estimations of their ideological positions) for years. But those who wanted to examine state legislators often had to assemble their roll call records painstakingly by hand or just rely on less dependable estimates. Boris Shor and Nolan McCarty managed to assemble roll call votes from all 99 state legislatures over the past two decades and place them, with the help of the NPAT survey, in a common space. This means that we can compare legislators across states and across time.

This seems an appropriate opportunity to commend those researchers who are making these and other valuable datasets available to other scholars and the public at large.

This is a huge advance, allowing us to examine how legislators behave under different or varying institutional rules. This dataset was at the heart of the forthcoming study on primary rules which shows that the type of primary a state has (open, closed, whatever) has no real impact on the partisanship of its elected officials. It can also give us real estimations of how congressional candidates will behave based on their previous work in state legislatures.

The American Ideology Project
Ever want to know just how liberal or conservative a state is? Or maybe a congressional district, a state legislative district, or even a city? Chris Tausanovitch and Christopher Warshaw pooled surveys of hundreds of thousands of Americans to create reliable estimates of districts' ideological predilections. This is a very useful collection, allowing us to measure the impact of redistricting, the re-election pressures faced by elected officials, and the difficulties confronting challengers, among other things.

Ideological Cartography
Political scientists often estimate the ideological leanings of politicians by examining their voting records. This makes sense—if you want to know what someone believes, look at how they vote—but has some notable limitations. For one thing, we can't estimate the ideological positions of candidates who have never served in a legislature.

Adam Bonica developed a method for calculating ideal points by studying who donates to candidates. This is extremely helpful to those who study elections, allowing us to see how much candidates' ideological positions matter in, say, primary elections, where many of the candidates might have no elected experience. (This study by Andrew Hall, for example, found that a party could lose around 12 points in the general election by nominating an ideological extremist instead of a moderate in the primary.) Bonica's data also allow us to estimate the ideological positions of donors themselves, so we can see just how ideological money has become.

A New Nation Votes
Is early American political history your thing? Then you'll want to check out this new collection of essentially every election held in the United States between 1789 and 1825, from president on down to town councils. Did you know James Madison beat James Monroe 133-2 in Louisa County, Virginia, in the 1790 contest to represent the state's 5th congressional district? You do now!

At any rate, as 2013 winds down, this seems an appropriate opportunity to commend those researchers who are making these and other valuable datasets available to other scholars and the public at large, and to encourage others to do so in the years to come.