Table S4.

Treatment effect estimates for study 2 voting preferences

Predictor variable	Presearch vote		Postsearch vote
	Coefficient	SE	Coefficient	SE
Intercept	−0.073	0.540	0.062	0.543
Sex
Female	0	Referent	0	Referent
Male	0.039	0.110	−0.135	0.119
Other	−0.430	0.922	−0.568	0.924
Race/ethnicity
White	0	Referent	0	Referent
Black	0.115	0.224	0.090	0.245
Hispanic	−0.435	0.235	−0.280	0.237
Asian	0.366	0.238	0.668	0.291*
Other	0.133	0.274	−0.072	0.291
Age group
18–24	0	Referent	0	Referent
25–44	−0.024	0.144	−0.083	0.157
45–64	0.241	0.184	0.029	0.200
65+	0.258	0.411	0.685	0.519
Education level
Less than ninth grade	0	Referent	0	Referent
Ninth to 12th grade	0.024	0.548	0.732	0.550
High school graduate	0.074	0.528	0.927	0.528
Bachelors	0.094	0.529	0.842	0.530
Advanced	−0.050	0.543	0.549	0.544

The presearch and postsearch columns report the estimate and variance for both treatment groups using classical regression poststratification. Data for sex, race/ethnicity, age group, and education level came from the 2010 US Census. Data on the number of people who identify their sex as “other” came from a 2011 Gallup study.

^*P < 0.05.