Anova and Nonparametric Tests

Applying ANOVA and Nonparametric Tests

The purpose of the simulation was to determine when it is feasible to use the ANOVA test over a nonparametric test while doing an analysis. In the case study done at Praxidike Systems, Team A has the task of researching the data provided, identify what the root cause or causes are, and provide recommendations for corrective action. The variables targeted for the research were customer satisfaction and productivity. Factors identified as affecting productivity were scope changes and type of project. Factors affecting client satisfaction was defects. In the simulation, we used the ANOVA test and the nonparametric Kruskal-Wallis test.

The three lessons that team A learned relative to ANOVA and nonparametric testing is that in order to use an ANOVA test you have to meet specific assumptions. Those assumptions would be that errors are random and independent of each other, each population has a normal distribution, and that all populations have the same variable. If we cannot determine if the population has a normal distribution, use a nonparametric test like the Kruskal-Wallis test to compare data for more than two populations. The simulation also taught us not to rush things and make the wrong conclusions; we need to take our time in analyzing the situation and maybe review the situation a couple of times before making a decision. It is easy to draw the wrong conclusion if certain facts are overlooked or by using the wrong methods for testing your analysis. It also helps if you have a clear understanding of what your choices are, and of what you have to decide.

In the workplace, such analysis tests will help break down situations like the one within the simulation. Problems can be analyzed faster and resolved quicker; these types of tests provides insight on issues that need to be broken down further, and also displays what it takes and what the involvement is to break down and find solutions to them. On the smaller...