Senate Voting on Abortion Legislation Over Two Decades: Testing a Reconstructed Partisanship Variable

Abstract

This paper argues that the ongoing debate over whether party or ideology is the primary determinant of legislative voting behavior has misdirected scholarly efforts. As conventionally modeled, analysts usually include a dichotomous party variable and an interval variable for ideology, but normally a dummy variable cannot outperform an interval variable when entered in multiple regressions. Because that methodology has been employed to assess voting on abortion in Congress, inevitably the results point to ideology as a strong predictor and party affiliation as, at best, a marginal influence. Also confounding this methodology is a recent concern that utilizing ADA scores as a proxy for ideology to predict voting behavior is a tautology, since votes are explaining votes. This study proposes a conceptual and methodological innovation by deriving a “reconstructed” partisanship variable for each Senator to determine how more partisan (conservative) Republicans and more partisan (liberal) Democrats voted on abortion legislation over the period 1973-1988. Besides partisanship, the eight regression models included three constituency influences (median family income, percent urban, and percent black) and the religious affiliation of each Senator (Catholic or non-Catholic). Overall the partisanship variable was the strongest predictor of senatorial voting on abortion bills and the results consistently showed that more partisan Democrats voted prochoice and more partisan Republicans voted pro-life.