A Test of the CIR 2013 Blog: The 287(g) Vote

By Tom K. Wong, tomkwong@ucsd.edu, @twong002

While we are still far from a vote on final passage of the comprehensive immigration reform (CIR) bill in the Senate, a vote yesterday in the House provides an early test of the models and predictions of the CIR 2013 Blog.

Yesterday, the House voted on an amendment to the Department of Homeland Security (DHS) appropriation bill that was introduced by Democratic Representative Jared Polis (CO-2). The amendment was related to the controversial 287(g) program, which promotes local law enforcement cooperation with federal immigration officials. Last week, I was asked to help count votes for and against the amendment.


The model predicted up to 185 yes votes on the Polis amendment. The actual votes was 180, which means that the model performed within 97% of the actual vote.

Taking a look at the actual roll call, we see that the model correctly predicted 94% of the actual votes cast – nearly 400 representatives were correctly predicted out of the 425 who voted.

92.3% of those predicted as “solid yes” votes actually voted yes. 92.3% of those predicted as “lean yes” votes actually voted yes. And 99.6% of those predicted as “solid no” votes actually voted no (Michelle Bachman was the lone “solid no” vote who voted yes). The model performed the worst with the “lean no” category. Only 31.5% of those who were predicted as “lean no” votes actually voted no.

Tom K. Wong, Ph.D., is an assistant professor of political science at UC San Diego. He is an expert on immigration politics and policy. He is a research associate at CCIS and beginning in fall 2013 he will be Director of the International Migration Studies Program at UC San Diego.