Improved Graphs for Calculating Sample Sizes When Comparing Two Independent Binomial Distributions

Abstract

Casagrande et al. (1978) compared the large sample sizes obtained from the exact method (Haseman 1978), the arcsine expression, the uncorrected x2 (Fleiss 1973), and a corrected x2 (Kramer-Greenhouse 1959). This comparison was made for different values of p, andp2 with cx = .05 and ,B-.90. They concluded that an average of the Fleiss and Kramer-Greerlhouse approximations would give a good estimate of the exact value from the Fisher-Irwin test,