Question
You have treadmill endurance test scores for 14 college studentstaken just before final exam period and a second set of scores forthe same students obtained on the afternoon of the last day offinals. You are pretty sure that the populations these scores aresampled from are highly skewed, and so you plan to test fordifferences in physical endurance using a nonparametric procedure.Which one should you use? a Wilcoxon T test
a two-way
X2 test
a Mann-Whitney Utest
a Friedman
X2 test
Explanation / Answer
The
Wilcoxon test for paired samples is thenon-parametric equivalent of the paired samples t-test. It shouldbe used when the sample data are not Normally distributed, and theycannot be transformed to a Normal distribution by means of alogarithmic transformation. The
Friedman test is a non-parametric testfor analyzing randomized complete block designs. It is an extensionof the sign test when there may be more than two treatments. (only 2 treatments in example)
Mann-Whitney U test for assessing whether twoindependent
samples of observations come fromthe same distribution (not what we need to test) The
two-way Chi Square test is a convenienttechnique for determining the significance of the differencebetween the frequencies of occurrence in two or more categorieswith two or more groups (not what we are testing for) I've never studied any of the mentioned tests here, and thisis what I learned from googling them. I believe that the Wilcoxon Ttest is the correct test.