Paired t-tests are widely used to compare differences in paired continuous outcomes. This test, however, does not apply to non-continuous outcomes. For binary outcomes,McNemar's test is the de facto method for com...Paired t-tests are widely used to compare differences in paired continuous outcomes. This test, however, does not apply to non-continuous outcomes. For binary outcomes,McNemar's test is the de facto method for comparing differences in paired binary variables. For count or frequency outcomes, there is not much discussion in the literature. The Wilcoxon signed-rank test is often used to assess differences in paired count outcomes. However, like the Mann-Whitney U test, or Wilcoxon rank-sum test for comparing two independent groups, the Wilcoxon signedrank test is generally not efficient(powerful). In this essay,we consider some alternative methods for comparing paired count outcomes and compare them with the Wilcoxon signed-rank test using Monte Carlo simulations.展开更多
基金partially supported by the National Institutes of Health,Grant UL1TR001442 of CTSA funding
文摘Paired t-tests are widely used to compare differences in paired continuous outcomes. This test, however, does not apply to non-continuous outcomes. For binary outcomes,McNemar's test is the de facto method for comparing differences in paired binary variables. For count or frequency outcomes, there is not much discussion in the literature. The Wilcoxon signed-rank test is often used to assess differences in paired count outcomes. However, like the Mann-Whitney U test, or Wilcoxon rank-sum test for comparing two independent groups, the Wilcoxon signedrank test is generally not efficient(powerful). In this essay,we consider some alternative methods for comparing paired count outcomes and compare them with the Wilcoxon signed-rank test using Monte Carlo simulations.
