ABSTRACT
This book offers randomization test macros for just a few designs, and you may by now be wondering why there are not more, or indeed why statistical software such as SPSS does not offer them. When we use the usual parametric and nonparametric statistical tests, we are making use of the fact that if the null hypothesis is true, it is possible to calculate the distribution of the test statistic. The distributions of most of the commonly used test statistics were tabulated decades ago. In order to see how probable is the value of the test statistic we obtained from our experiment, we only have to consult a table. In fact, when using software such as SPSS, we do not even need to do that because the probability under the null hypothesis of a value at least as extreme as the one we obtained is given with the test statistic. In contrast, to use a randomization test, we have to find the values in the reference set ourselves. Whether we list and calculate every value in the reference set, or use a sampling method as in our macros, obtaining the reference set for comparison with our experimental value is not a process that can be fully generalized. So each design needs its own macro or other software. If you want to use a design that does not fit one of our macros, you could see if there is other software that will support the analysis of your chosen design. In the “Other Sources of Software for Randomization Tests” section of this chapter, we briefly list some sources of software other than our macros. Another option is to modify one of our macros to fit your design. The penultimate section in this chapter explains how you can do this.
