Party Control and Perverse Effects in Majority-Minority Districting: Replication Challenges When Using DW-NOMINATE

Every 10 years, states set about redrawing the lines of their Congressional districts. Scholars in political science have long been interested in the strategic behavior and representational outcomes of this process. While majority-minority districts are intended to provide a constraint on strategic party behavior in order to ensure substantive representation of minority interests, researchers have noted a perverse effect that results in potentially less-representative political outcomes. In 2003, Kenneth Shotts, David Lublin, and D. Stephen Voss debated the veracity of the perverse effects claim, but Shotts' critique was missing a key interaction between partisanship and the liberalizing effect of majority-minority districts. In the course of performing this necessary extension on Shotts' work, we found that our results have an unexpected, and important, methodological implication for Congress scholars. Specifically, we were unable to replicate his results precisely due to sublte changes in DW-NOMINATE estimates that result from periodic updating of the database. Further-more, after substantially expanding the dataset, we continue to find the same null results and the evidence supporting the interaction is statistically ambiguous. Though these null results do not prove or disprove the perverse-effects hypothesis, they do undermine Shotts' evidence of a liberalizing effect of majority-minority districting. While we lack sufficient precision to estimate whether majority-minority districting has a positive, negative, or truly no effect on minority representation (and the conditional effect of party control), it is more concerning that small changes to DW-NOMINATE would prevent the replication of these past results, given the abundance of studies that use it to measure legislator ideology.