Table 1.

Effect of voter and voting treatments on 2014 primary election turnout

Sample	Proportion voting, voter treatment	Proportion voting, voting treatment	Difference of proportions (voter − voting) [SE]	Regression estimate of difference (voter − voting) [SE]	No. of observations (voter; voting)
Entire sample	0.301	0.311	−0.010 [0.014]	−0.004 [0.011]	2,236; 2,232
State: Michigan	0.234	0.242	−0.008 [0.019]	−0.007 [0.015]	1,041; 1,015
State: Missouri	0.333	0.341	−0.008 [0.028]	−0.005 [0.023]	577; 563
State: Tennessee	0.383	0.393	−0.009 [0.027]	0.000 [0.021]	618; 654
No competitive house primary	0.321	0.339	−0.018 [0.018]	−0.011 [0.015]	1,307; 1,305
Either house primary competitive	0.273	0.273	0.000 [0.021]	0.004 [0.016]	929; 927
Ever voters	0.326	0.338	−0.013 [0.015]	−0.004 [0.012]	2,047; 2,033
Have voted in primary	0.596	0.632	−0.037 [0.024]	−0.014 [0.021]	848; 832
Have voted but never in primary	0.135	0.135	0.000 [0.014]	0.003 [0.013]	1,199; 1,201
No prior history of voting	0.032	0.035	−0.003 [0.018]	−0.003 [0.018]	189; 199
Predicted turnout >70%	0.764	0.781	−0.017 [0.030]	−0.007 [0.027]	377; 407

The estimates in column 5 were generated from regression models including strata (strata × vote history × district competitiveness) fixed effects and state interacted with indicators for age, year of registration, sex, race/ethnicity, and the number of times voted in general, primary, and special elections (complete model results are reported in Table S3). No differences in proportions or regression estimates are statistically significant (P < 0.05).