Published online by Cambridge University Press: 23 March 2020
According to the burgeoning voter mobilization field experiments literature, impersonal contact has little-to-no effect. A conversation with a computer is, by definition, an impersonal interaction. However, chatbots are form of communication wherein a computer imitates conversations by interacting using natural language. Human-to-chatbot interactions are often perceived as similar to human-to-human interactions. How ubiquitous are chatbots? Just ask Siri, Cortana, Bixby, or Alexa, or ask for help on any e-commerce website (these are each chatbots). A simple voter mobilization treatment reminding users of a political chatbot (Resistbot) to vote and providing information about polling locations and hours increased turnout by 1.8 percentage points in a 2019 election. The results replicate previously unpublished field experiments by Resistbot in 2018 that found smaller but statistically significant increases in turnout.
I am grateful to Jason Putorti of Resistbot for sharing the data from the experiments with unrestricted right to publish. This research received financial support from Resistbot in the form of (1) access to data and (2) a small payment (<$2500) to the primary investigator as compensation for time to analyze the data and prepare a memo for Resistbot funders. The contract between the primary investigator and Resistbot guaranteed unrestricted publication rights using the data. Resistbot did not have a right to review, nor did it review, the manuscript prior to submission. I thank the reviewers and Associate Editor for valuable comments to improve the paper. All analysis and interpretation are my responsibility. The data, code, and any additional materials required to replicate all analyses in this article are available at the Journal of Experimental Political Science Dataverse within the Harvard Dataverse Network at https://doi.org/10.7910/DVN/BBPLQH.