An Alternative Interpretation of Freedman's Nonresponse Case Study
重新分析了Freedman(1986)的应答-无应答数据，发现存在无应答偏差，并指出协变量在此案例中未能有效比较两组，建议通过后续抽样获取更强证据。
Further analysis of the response–nonresponse data presented by Freedman (1986) shows evidence of a nonresponse bias. Covariates can be useful in comparing respondents with nonrespondents; however, we argue that they were not effective in this case. Follow-up sampling would provide stronger evidence concerning the similarity of the two groups.