POWER COMPARISON OF SOME PARAMETRIC AND NON-PARAMETRIC TESTS

Abstract

In this study power of statistical tests is used as a criterion for selection of best method.  Experiments were subjected to both parametric and non-parametric methods of analysis; large and small samples were considered in each case. It was found that for large samples, the parametric tests have an average power of 80.57% and the non-parametric has 77.74% for small samples, the parametric tests gave an average power of 52.43% while non-parametric tests recorded 35.76%. When both large and small samples were combined, the parametric tests gave an average power of 66.5% while the non-parametric tests recorded 56.85%. It could be concluded that parametric tests are generally more powerful than their non-parametric counterparts and also large samples are more powerful than small samples.