Working Paper: CEPR ID: DP13991
Authors: Joo Pereira dos Santos; Jos Tavares; Pedro C. Vicente
Abstract: We report on a large-scale field experiment to assess ATMs (automatic teller machines) capacity to “get out the vote”. This is a heretofore unexploited method. Our experimental design used the universe of functioning ATMs in Portugal. We randomly selected a set of treatment civil parishes, where a civic message took over the totality of ad time, which we compare with a set of control areas. The campaign we follow was active for three days before and during the 2017 local elections. Although we do not achieve statistical significance on a stable but small average treatment effect, when we consider the intensity of treatment, results show a statistically significant increase in the likelihood of voting. Placebo tests using turnout rates in previous elections strengthen our interpretation. We ran a post-treatment survey around ATMs located in two neighbouring civil parishes, one treated, the other not. We found a sizeable difference in recall.
Keywords: voter mobilization; ATMs; Portugal; local elections
JEL Codes: C93; D72; H70
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| observational studies (C90) | positive correlations between campaign expenditures and voter turnout (K16) |
| ATM messaging (L96) | voter turnout (K16) |
| treatment intensity (C32) | voter turnout (K16) |
| placebo tests (C90) | no significant impact on turnout rates from previous elections (K16) |
| post-treatment survey (C83) | difference in recall of treatment messages (C90) |