AN ANALYSIS OF STATE REFERENDA
Richard E. Matland
Loyola University Chicago
Paper presented to the University of Wisconsin, Department of Political Science American Politics Workshop, March 21, 2011
Between 1998 and 2009 thirty-four states held referenda on prohibiting or permitting same-sex marriages. All but one resulted in prohibition. The outlier was Arizona, which in 2006 narrowly defeated Proposition 107 prohibiting same sex marriage (48%-52%). Two years later, however, Arizonans switched and voted 56% to 44% to prohibit same-sex marriages. All told, voters directly approved constitutional prohibitions in thirty-one states, of which twelve were citizen-initiated referenda and nineteen referenda were authorized by state legislatures. This paper presents an analysis of voting outcomes on these statewide referendums across the 34 states that have had votes during this time. Our units of analysis are the 2,344 counties in thirty different states where referenda voting occurred. We analyze an array of socio-economic, political, and institutional variables to explain the pattern of opposition to same-sex marriage. This paper builds on existing research of political scientists and sociologists by (1) integrating the salient variables from previous studies, (2) relying on the universe of referenda rather than a subset, (3) adding state-level variables and (4) including “structural” variables associated with the referendum process. In sum, we believe our study is the most comprehensive analysis of this research question to date. We also believe several of the new variables we include are significant contributions as they focus on the pivotal impact of voter “turnout” in explaining the level of opposition to same-sex marriage in the American states and the impact that legislative language has on the outcomes we evaluate. Literature Review and Research Hypotheses
Public opinion surveys have shown an increasing acceptance of gays throughout American society over the past 30 years. With the dramatic repeal of Don’t Ask Don’t Tell in December 2010 yet another significant barrier has fallen. Yet on the issue of same sex marriage the LGBT coalition has seen itself repeatedly beaten and often soundly at the state level. Today the controversy over same-sex marriage is one of the major battlegrounds in the “culture wars” that divide America (Hunter, 1991 but see Fiorina, 2005) This controversy has provoked scholarly interest (Rimmerman and Wilcox, 2007; Segura, 2005; Soule, 2004), particularly during the 2004 election when eleven same-sex referenda were on the ballot (Lewis, 2005; Hillygus and Shields, 2005). One claim is that the heightened voter turnout for president influenced those referenda outcomes, but there has been no systematic analysis of the impact of voter turnout on voting against same-sex marriage for the universe of cases. That research question primarily drives our analysis, though we also test a variety of hypotheses drawn from sociology, religion and politics, and the literature on referendum voting and electoral behavior. There are two previous studies of referenda voting on same-sex marriage at the county level, one by two political scientists (Lofton and Haider-Markel, 2007) and one from a sociological perspective (McVeigh and Diaz, 2009). Lofton and Haider-Markel (2007) analyzed county vote in the 2004 referenda and found the following county attributes correlated positively with a yes vote on the referenda to ban same-sex marriage (i.e. opposition to same-sex marriage): Republican partisanship, Protestant fundamentalism, higher and non-urban proportion of residents. The following county attributes correlated negatively with a yes vote on such referenda: voter turnout, proportion of gay households, male-to-female ratios, and proportion of college educated persons. McVeigh and Diaz (2009)...