Monte Carlo Researchers Technique to Evaluate the Welch as the Default

• In Event: Poster Session for Quantitative Methods and Statistical Theory

Abstract

This study illustrates, using Monte Carlo simulation, that the Welch's t test should usually be preferred over the pooled independent t test (Conditional t test) when sample sizes are unequal. Even small differences in variances lead to inflated Type I errors using Conditional t tests, this study introduces an alternative solution to a case where variances are unequal with two and three groups. Monte Carlo strategies were used to examine the criterion for the Conditional Decision Rule for testing statistically significant differences between means. The independent groups t test is one of the most widely used statistics, however, in many situations, the assumption of nonnormality and homogeneity of variance is not met, and an invalid inferential statistical test may result.

Authors

• Elizabeth Ottie Ayisi, Ohio University

• Edward Matovu, Ohio University

• Nina Adjanin, Ohio University

• Gordon P. Brooks, Ohio University