Testing Non-Nil Null Hypotheses Using Resampling Stats Add-In for Microsoft Excel

J. Frass and I. Newman (2000) proposed a hypothesis testing procedure that incorporated the following three key elements: (1) the establishment of a practical significance value; (2) the construction of a non-nil null hypothesis that incorporated the practical significance value; and (3) statistical testing of the non-nil null hypothesis with a randomization test. One of the difficulties researchers may encounter with this testing procedure is the implementation of the randomization test. This paper describes, through use of an example, how researchers can conduct a randomization test with relative ease with the use of the computer software Resampling Stats Add-in for Excel. In the final section of the paper, the randomization test results for the example data are compared to independent samples t-test results. The outcome of this comparison suggests that future investigation of the relative results of the two types of statistical tests may be beneficial to researchers. An appendix contains the computer program for the randomization test. (Contains 44 references.) (Author/SLD)