Online Program

Friday, February 21
PS2 Poster Session II & Refreshments Fri, Feb 21, 4:45 PM - 6:15 PM
Bayshore II-IV

A Simulation Study to Compare the Performance of Independent Means t-test and Alternatives in Terms of Type I Error and Statistical Power (302846)

Aarti Bellara, University of South Florida 
Yi-Hsin Chen, University of South Florida 
Anh P. Kellermann, University of South Florida 
Eun Sook Kim, University of South Florida 
Jeffrey D. Kromrey, University of South Florida 
*Diep Thi Nguyen, University of South Florida 
Patricia Rodriguez de Gil, University of South Florida 

Keywords: Type I error control, statistical power, parametric tests, Satterthwaite’s approximate test, Conditional t-test

This study explored the performance of the independent t-test and alternatives (i.e., Satterthwaite’s t- test and the conditional t-test) under various conditions. The simulated factors examined in this study included total sample size, sample size ratio between groups, variance ratio between populations, effect size for mean difference between populations, alpha set for testing treatment effects, alpha set for testing homogeneity assumption, and population distributions with varying kurtosis and skewness values. The crossed factorial design provided a total of 176,400 conditions. This study confirmed that the independent means t-test is robust to violations of the normality assumption when two population variances are equal and robust to violations of the homogeneity assumption when the sample sizes are equal. The Satterthwaite t-test provided adequate type I error control in nearly all conditions and the conditional t-test evidenced notable improvement in type I error control relative to the independent means t-test as the level of alpha for the test of variances increased. The results also provided guidance in selecting proper test with nonnormal, heteroscedastic populations.