Forecasting Ideology in the 113th Congress (2013-14)
By Adam Bonica and Howard Rosenthal
We estimated the liberal-conservative ideology of candidates in the 2012 elections by using contributions they received through November 6, 2012 from individual donors and issue oriented PACs. Our estimates, which also use data beginning in 1980, reflect tens of millions of individual contributions reported by candidates to the Federal Election Commission. We processed this data with an algorithm that matches contributor names and addresses across candidates. The result is a giant spreadsheet like array where the rows are candidates and the columns individual contributors. The estimation that processes this array is somewhat analogous to the well-known DW-NOMINATE estimation of liberal-conservative ideology from roll call voting. For contributions, the contributors are similar to legislators and the legislators are similar to roll calls. The amount of money given influences the estimation. These amounts are, in fact, more informative than the simple “Yea-Nay” choices of the roll call votes.
Using contributions through any period before an election, we can then see whether contribution ideology matches up with roll call ideology in the ensuing Congress. We have used the estimated equation that best fits actual DW-NOMINATE scores for the 96th-112th Congresses as the equation for predicting the 113th Congress.
Polarization is measured as the distance between the average Republican and the average Democrat. Scholars have typically put the independents Bernie Sanders and Joe Lieberman in with the Democrats. For our predictions for the 113th Congress, we have put Sanders and Angus King in with the Democrats. King’s contributions suggest he will vote as a liberal similar to Diane Feinstein (CA) or Patrick Leahy (VT).
Since CF scores and DW-NOMINATE scores are computed independently they are not directly comparable. We normalized both such that they have a standard deviation of 1.0. The normalized DW-NOMINATE scores remain very comparable to the original ones. The normalization is useful for graphical presentation. It is not necessary for forecasting DW-NOMINATE scores for new members.
The two line graphs below shows that the estimates, applied to all elections starting in 1980, are highly accurate. The very end of the predicted line shows the 113th Congress forecast, still more polarized than the previous record attained by the 112th Congress. Two charts follow the line graphs that show the observed and forecast distribution of ideology for the 99th (1985-85) and 113th (2013-14) Houses. The huge gap between the parties is apparent. We may hope that senators and representatives reach across the aisle but the aisle is now as wide as the Grand Canyon.