Replication Data: Officer Diversity May Reduce Black Americans’ Fear of the Police

This dataset includes responses to two survey experiments testing the effects of officer diversity, which were conducted with a national sample (N = 1,100). The survey was fielded by YouGov in the spring of 2022 (between April 21 and May 2). For our experiments, YouGov constructed two synthetic sampling frames (SSF) via stratified sampling from the 2019 American Community Survey, which were used to select two matched (on gender, age, and education) samples of opt-in panelists: a general population sample (N = 650) and a large oversample of Black Americans (N = 450). (The general population sample was also matched on race.) Using propensity scoring based on region and the matching variables, both samples were then weighted to their respective SSFs, after which the weights were post-stratified on 2016 and 2020 Presidential vote choice. The purpose of the oversample was to yield (after combining Black respondents in the oversample with those in the general population sample) similarly sized analytic samples of Black and non-Black Americans. Per this sampling design, we estimated the models for the experiments separately for Black and non-Black respondents. For the main analysis, we applied the provided sampling weights. (NOTE: The original files were uploaded in Stata-12 version.)

本数据集包含两项检验公职人员多样性影响的调查实验的应答数据，两项实验基于全国样本（N=1100）开展。本次调查由调研机构YouGov于2022年春季（4月21日至5月2日期间）实施。在本次实验中，YouGov通过对2019年美国社区调查（American Community Survey）进行分层抽样，构建了两个合成抽样框架（Synthetic Sampling Frame，SSF），以此选取两组在性别、年龄与教育水平上匹配的自愿参与面板受访者样本：其一为普通人群样本（N=650），其二为美国黑人受访者的超额抽样样本（N=450）。（注：普通人群样本同时在种族维度上完成了匹配。）随后基于地区与匹配变量采用倾向得分法，对两个样本分别按各自的合成抽样框架进行加权；完成上述加权后，再基于2016年与2020年美国总统大选的投票选择对权重进行后分层调整。本次超额抽样的目的在于，将超额抽样样本中的黑人受访者与普通人群样本中的黑人受访者合并后，可获得规模相仿的黑人与非黑人美国受访者分析样本。基于此抽样设计，我们分别针对黑人与非黑人受访者估算了实验模型。主分析环节我们采用了配套提供的抽样权重。（注：原始文件以Stata-12版本上传。）