Do and log files for: Examining Voting Spillover Effects of Text Message Reminders

We present the results from a pre-registered and well-powered ($N>3,000,000$) text-message get-out-the-vote (GOTV) experiment, conducted in conjunction with the 2019 European Parliament election in Sweden. Our findings suggest that a simple text-message increases the likelihood of voting by 0.3 percentage points. Half of this effect spills over to the untreated household members, while the spillover effects within workplaces are close to zero. Subsequent analysis reveals that the direct treatment effect is noticeably stronger among individuals with below-average voting propensities. Interestingly, within this same group, the household spillover effects are significantly negative. We speculate and provide some indirect evidence, that these negative spillover effects may stem from the text-message reminder influencing the behavior of voters who were already motivated to vote. Above all, we propose that an increase in early voting, as opposed to voting on Election Day, among treated individuals may weaken the mechanisms thought to explain spillover effects, since voters are less likely to bring their family members with them when voting early. This is the do-file used to produce all our results, together with the corresponding log file. Unfortunately we are not allowed to share the data necessary for the replication.