I'd Be Surprisingly Good for You: Political Information and Network Effects

We use a randomized experiment to assess the impact of a political information campaign on voters who were directly exposed to it and on other voters in their neighborhood (the direct and indirect effects, respectively). During the 2023 presidential elections in Argentina, Javier Milei proposed, among other things, replacing free public education with a voucher system. Following his success in the August 2023 PASO primaries, we sent leaflets providing fact-checking information on the consequences of Milei's policy proposals to a subset of voters in Salta Province, where Milei had the largest vote share in the PASO. Argentina's reporting system and our experimental design enable us to estimate both the direct and indirect effects of the campaign. The campaign reduced support among leaflet recipients (direct effect), but it increased support for him among their neighbors (indirect effect), with the indirect effect strong enough to boost Milei's overall performance. These outcomes persisted in the runoff elections. We replicated the experiment with a different sample of subjects in the runoff and obtained similar results. The paper makes a methodological contribution by estimating within-neighborhood indirect effects using official election results. Our study shows that direct and indirect effects may have opposite signs, and it has implications for how and why information campaigns may backfire.