Research Interests:
Research Interests: Mathematics and Jar
Research Interests:
Research Interests:
Research Interests:
The Freeman-Tukey double-arcsine transformation is used to transform binomial or Poisson data so that they correspond with probabilities under the standard normal. This permits use of the normal distribution for significance testing on... more
The Freeman-Tukey double-arcsine transformation is used to transform binomial or Poisson data so that they correspond with probabilities under the standard normal. This permits use of the normal distribution for significance testing on proportions. Tukey’s HSD and other “post hoc” multiple comparison tests have traditionally been used to test differences among groups of means. The double-arcsine transformation allows for the use of these to test differences among proportions in 2xc contingency tables. This paper presents a Base SAS macro that will run Tukey’s HSD, LSD, Bonferroni, Dunn-Sidak and Scheffé tests on groups of proportions having a 2xc structure. Simulation testing showed that each of these tests produced more conservative results and protected family-wise error rates much closer to alpha risk than pairwise χ tests of association on the same proportions. The macro permits inputs as a 2xc counts table or as proportions and their associated sample sizes.