Replication Data for: A 2 million person, campaign-wide field experiment shows how digital advertising affects voter turnout

We present the results of a large, $8.9 million campaign-wide field experiment, conducted among 2 million moderate and low-information “persuadable” voters in five battleground states during the 2020 US Presidential election. Treatment group subjects were exposed to an eight-month-long advertising program delivered via social media, designed to persuade people to vote against Donald Trump and for Joe Biden. We found no evidence the program increased or decreased turnout on average. We find evidence of differential turnout effects by modeled level of Trump support: the campaign increased voting among Biden leaners by 0.4 percentage points (SE: 0.2pp) and decreased voting among Trump leaners by 0.3 percentage points (SE: 0.3pp), for a difference-in-CATES of 0.7 points that is just distinguishable from zero (t(1035571) = −2.09, p = 0.036, DIC = 0.7 points, 95% CI = [−0.014, −0.00]). An important but exploratory finding is that the strongest differential effects appear in early voting data, which may inform future work on early campaigning in a post-COVID electoral environment. Our results indicate that differential mobilization effects of even large digital advertising campaigns in presidential elections are likely to be modest.

本研究呈现了一项规模宏大、总投入达890万美元的全竞选范围现场实验结果。该实验于2020年美国总统大选期间，在五个摇摆州的200万名持温和立场且信息获取不足的「可说服型选民（persuadable voters）」中开展。实验组选民接收了为期八个月的社交媒体广告投放计划，该计划旨在说服选民投票反对唐纳德·特朗普，支持乔·拜登。研究未发现该计划整体上提升或降低选民投票率的证据。但研究发现，基于建模得到的特朗普支持倾向水平，存在差异化的投票率效应：该竞选活动使倾向拜登的选民投票率提升0.4个百分点（标准误：0.2个百分点），使倾向特朗普的选民投票率降低0.3个百分点（标准误：0.3个百分点），对应的条件平均处理效应差异（difference-in-CATES, CATES）为0.7个百分点，该结果显著异于零（t(1035571) = −2.09, p = 0.036, 偏差信息准则（DIC）= 0.7个百分点，95%置信区间（95% CI）= [−0.014, −0.00]）。一项虽具探索性但意义重大的发现是，最强的差异化效应出现在提前投票数据中，这一结果可为后新冠疫情选举环境下的早期竞选研究提供参考。本研究结果表明，美国总统大选中，即便大规模数字化广告竞选活动，其带来的差异化选民动员效应也大概率较为有限。