This chapter introduces a general technique, evidence factors, in its simplest application, namely, the evaluation of the new information provided by a second control group. The rationale underlying the use of multiple control groups is briefly reviewed. A study has two evidence factors if it permits two essentially independent tests of hypotheses about treatment effects, where the unmeasured biases that affect one test do not affect the other, even though both tests may be biased. Because the two tests are essentially independent, they may be combined using meta-analytic tools as if they came from unrelated studies of unrelated data, even though they actually reanalyze the same data from two orthogonal perspectives. A good test for the second evidence factor is found with the aid of design sensitivity and Bahadur efficiency.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Evidence Factors for Two Control Groups

  • Paul R. Rosenbaum

摘要

This chapter introduces a general technique, evidence factors, in its simplest application, namely, the evaluation of the new information provided by a second control group. The rationale underlying the use of multiple control groups is briefly reviewed. A study has two evidence factors if it permits two essentially independent tests of hypotheses about treatment effects, where the unmeasured biases that affect one test do not affect the other, even though both tests may be biased. Because the two tests are essentially independent, they may be combined using meta-analytic tools as if they came from unrelated studies of unrelated data, even though they actually reanalyze the same data from two orthogonal perspectives. A good test for the second evidence factor is found with the aid of design sensitivity and Bahadur efficiency.