Test Power in Comparison Difference between Two Independent Proportions
Test power, sample size, simulation, comparing two proportions
In this study, the effect of differences between population proportions (effect size) and relationships between sample sizes on test power in comparing two independent proportions was investigated. At the end of 50 000 simulation trials it was observed that increases in sample size and population proportion differences increased test power while decreasing the sample size and population proportion differences decreased test power. In the case of studies with equal sample sizes, sufficient test power level (80.0 %) was reached with 60, 90, 150 and 350 observations when d= 0.25, 0.20, 0.15 and 0.10, respectively. On the other hand, it could not reach sufficient power level even for the extremely large sample size (500 observations) when d=0.05. Results of this study showed that the inequality in sample sizes or relations between sample sizes (n2=r.n1 or n1=r.n2) affect the test power. However, total number of observations may be more effective on the test power rather than inequality in sample sizes.