Apply

Applying

Applying ANOVA and Nonparametric Tests Simulation

Crystal Beauchamp

RES/342

December 17, 2011

Professor Feravich

Applying ANOVA and Nonparametric Tests Simulation

Praxidike Systems, a software solutions company needed to regulate the reason for the postponement of 10% of its projects, low productivity, and low customer satisfaction. The Kruskal-Wallis and ANOVA tests were used to identify problem areas and recommendations for improvement were made based on the findings. With ANOVA, the following assumptions made: the population has a normal distribution; all errors are independent; and population variance is the same. A separate test must be conducted to determine if the population has a normal distribution - the chi-square test checks for goodness of fit. Conversely, the Kruskal-Wallis test does not assume normal distribution, but the data is on an ordinal scale.

Lessons Learned

During the ANOVA and Nonparametric tests simulation, I learned many lessons. The main assumptions in the use of the ANOVA test are 1) errors are random and independent, 2) the population as a normal distribution, and 3) all populations have the same variance. The ANOVA test can be used to test for a significant difference between means as well. With the Nonparametric test, there are no assumptions made on population distribution; it uses a magnitude based ranking. During this simulation, I made mistakes by choosing the incorrect test. I learned that it is important to choose the correct test based on the characteristics of the data. I also made some mistakes that were based on misread results of the test. In making recommendations based on an analysis of data, it is vital to read the data results correctly to begin with. If the results are misinterpreted, one cannot make an accurate recommendation to correct the business problem. Tests help us to identify where, and if, valuable resources should be used to correct a problem. Tests can be...