Version 3

Weights were created to ensure proper representation of the U.S. population. The study team created base weights that considered the original probabilities of selection. The study team also adjusted for non-resolution and non-response at the screening and interview stages. Finally, the study team implemented raking adjustments to control-totals from the Current Population Survey based on age, gender, race-ethnicity, educational attainment, and Census division (region). After the weights were calculated they were combined by assigning a factor to each sample type based on the total number of interviews contributed by each source. The study team then implemented an additional round of raking adjustments using the same control totals as in previous steps. In data analysis, the unweighted data were used for the cluster analytical work that generated the typology. The weighted data were used to describe group differences on the demographic, political and health status measures and for all other data analysis work including the preparation of sample-wide frequency distributions.本数据集构建权重的初衷是确保样本能够精准代表美国总人口。研究团队首先构建了基础权重，该权重纳入了初始抽样的入选概率；同时针对筛查与访谈阶段出现的未完成访问与无应答情况进行了加权校正。最后，研究团队依据年龄、性别、种族族裔、受教育程度及人口普查分区，参照美国当前人口调查（Current Population Survey）的控制总量，实施了raking加权校准调整。权重计算完成后，研究团队根据各数据来源贡献的总访谈样本量，为每类样本分配权重因子，以此合并各类样本的权重。随后，研究团队沿用此前步骤的控制总量，开展了第二轮raking加权校准调整。在数据分析环节，未加权数据被用于构建分类体系的聚类分析工作；而加权数据则用于描述人口统计学、政治立场与健康状况维度上的群体差异，同时也用于其余所有数据分析工作，包括生成全样本频次分布表。