Statistical significance and practical significance can be considered jointly through the use of non-nil null hypotheses that are based on values deemed to be practically significant. When examining differences between the means of two groups, researchers can use a randomization test or an independent t test. The issue addressed in this paper is whether Type I error rates produced by independent t tests of group means are impacted by the use of non-nil null hypotheses. The results of this Monte Carlo study suggest the frequencies of Type I error rates produced by independent t tests of group means for non-nil hypotheses are comparable to those recorded for nil null hypotheses. These results should provide further encouragement for researchers to use non-nil hypotheses as a means of considering statistical and practical significance jointly. (Contains 5 tables and 15 references.) (Author/SLD)
