The Williams Institute

Recent polls report majorities of the public supporting marriage recognition for same-sex couples. These reports are not uniform, with some polling organizations still reporting less than a majority in favor. I examine variation in these results using meta-analysis to examine variation among organizations (norg=21) and question wordings (nqw=36) . I also examine direct effects that explain variation based on question framing from 1996 to 2014 (npoll=138) . The results show that after accounting for this variation, a majority of the public supports same-sex marriage in 2014, and any deviations from that estimate can be explained by question framing and organizational “house” effects. The results indicate that questions that focus on the issue of “same-sex marriage” garner less support than questions on legal recognition of marriages for same-sex couples; this finding suggests that framing accounts for why some polls do not report majorities in favor.