Week 9.2 discussion response
Write a response to each discussion post.
Nonparametric tests are tests that are used to analyze nominal and ordinal data, skewed data, or when the conditions or assumptions for parametric tests are not able to be concluded (Erford, 2015). Normally, parametric tests are more powerful than nonparametric tests because they are able to use the raw data, therefore not having to do transformations of data. When data is transformed it can create the risk of losing some of the interaction or main effect that is present when using the real data (Erford, 2015). That being said, nonparametric tests show advantages in statistical power over parametric tests when the data is skewed. Some examples of nonparametric tests are the Kolmogorov-Smirnov, Wilcoxen Rank Sum, Mann-Whitney U, and Kruskal-Wallis tests (Erford, 2015). To better describe, consider the Wilcoxen Rank Sum. This is a nonparametric test that would be used instead of the t-test when the population is not assumed to be normally distributed and when the researcher is interested in comparing two samples to determine whether the mean ranks of the populations differ (Erford, 201
Nonparametric test is mostly used to determine the normal and ordinal data to ensure if the outlier was landed in the correct place (Erford, 2015 p.386). However, parametric test is more powerful compared to nonparametric test due to the fact that it more better of in fingering out the significant statistical differences among the group populations. For instance, if the graphs were too curvy, the data will twists, and the outcome would look different because it doesn’t meet the prediction for parametric test, for instance, histogram or boxplot. When the parametric tests is Residual, the score minus the predicted values group means. Whereas, parametric statistical assume the variables (residuals) are normally distributed (Erford, 2015 p. 393).