Choose Tests (Independent-Samples Nonparametric Tests)
These settings specify the tests to be performed on the fields specified on the Fields tab.
Automatically choose the tests based on the data. This setting applies the Mann-Whitney U test to data with 2 groups, or the Kruskal-Wallis 1-way ANOVA to data with k groups.
Customize tests. This setting allows you to choose specific tests to be performed.
-
Compare Distributions across Groups. These produce independent-samples tests of whether the
samples are from the same population.
Mann-Whitney U (2 samples) uses the rank of each case to test whether the groups are drawn from the same population. The first value in ascending order of the grouping field defines the first group and the second defines the second group. If the grouping field has more than two values, this test is not produced.
Kolmogorov-Smirnov (2 samples) is sensitive to any difference in median, dispersion, skewness, and so forth, between the two distributions. If the grouping field has more than two values, this test is not produced.
Test sequence for randomness (Wald-Wolfowitz for 2 samples) produces a runs test with group membership as the criterion. If the grouping field has more than two values, this test is not produced.
Kruskal-Wallis 1-way ANOVA (k samples) is an extension of the Mann-Whitney U test and the nonparametric analog of one-way analysis of variance. You can optionally request multiple comparisons of the k samples, either all pairwise multiple comparisons or stepwise step-down comparisons.
Test for ordered alternatives (Jonckheere-Terpstra for k samples) is a more powerful alternative to Kruskal-Wallis when the k samples have a natural ordering. For example, the k populations might represent k increasing temperatures. The hypothesis that different temperatures produce the same response distribution is tested against the alternative that as the temperature increases, the magnitude of the response increases. Here, the alternative hypothesis is ordered; therefore, Jonckheere-Terpstra is the most appropriate test to use. Smallest to largest specifies the alternative hypothesis that the location parameter of the first group is less than or equal to the second, which is less than or equal to the third, and so on. Largest to smallest specifies the alternative hypothesis that the location parameter of the first group is greater than or equal to the second, which is greater than or equal to the third, and so on. For both options, the alternative hypothesis also assumes that the locations are not all equal. You can optionally request multiple comparisons of the k samples, either All pairwise multiple comparisons or Stepwise step-down comparisons.
- Compare Ranges across Groups. This produces an independent-samples tests of whether the samples have the same range. Moses extreme reaction (2 samples) tests a control group versus a comparison group. The first value in ascending order of the grouping field defines the control group and the second defines the comparison group. If the grouping field has more than two values, this test is not produced.
- Compare Medians across Groups. This produces an independent-samples tests of whether the samples have the same median. Median test (k samples) can use either the pooled sample median (calculated across all records in the dataset) or a custom value as the hypothesized median. You can optionally request multiple comparisons of the k samples, either All pairwise multiple comparisons or Stepwise step-down comparisons.
- Estimate Confidence Intervals across Groups. Hodges-Lehman estimate (2 samples) produces an independent samples estimate and confidence interval for the difference in the medians of two groups. If the grouping field has more than two values, this test is not produced.