Kruskal-Wallis ANOVA and Median Test - Quick Tab

The Quick tab of the Kruskal-Wallis ANOVA and Median Test dialog box contains the options described here.

Variables. Click the Variables button to display the standard variable selection dialog box, in which you select the independent (grouping) variable and the dependent variable(s).

Codes. Click the Codes button to display the Select codes for grouping variable dialog box, in which you can specify the codes that were used in the grouping variable (in order to uniquely identify the group membership of each case). Note that the Codes button is available only after variables have been selected.

Summary: Kruskal-Wallis ANOVA & Median test. Click this button to display either the standard variable selection dialog box or the results of the Kruskal-Wallis ANOVA and Median test (if you have already selected the variables). See Nonparametric Statistic Notes - Kruskal-Wallis ANOVA by Ranks and Median Test for a detailed description of this nonparametric test.

Multiple comparisons of mean ranks for all groups. Click this button to compute post-hoc comparisons of mean ranks of all pairs of groups (see Siegel & Castellan, 1988, p. 213-215). Two results spreadsheets will be displayed for each dependent variable. The first spreadsheet shows the z values for each comparison between groups u and v, computed as:

where R-bar denotes the average ranks for the two groups, and nu and nv are the number of observations in the two groups (u and v). Note that this spreadsheet will also show the mean ranks in the second rows of the respective column headers (denoted as R:.. ).

The header of the spreadsheet will identify these values as z' (z-prime), because depending on the type of comparisons you want to perform, these z values will correspond to different p-values (significance levels). Specifically, the second results spreadsheet will show the p-values (two-sided significance levels with a Bonferroni adjustment) associated with each comparison, computed as:

p=p(z')*k*(k-1)

where p(z') = 2 * (1-Pr(Z<z')), and k stands for the total number of groups in the comparison.

Note that you can also compute the post-hoc probabilities for comparing k-1 groups against a control group by applying:

p=p(z')*2*(k-1)

Refer to Siegel and Castellan (1988, equations 8.7, 8.8) for additional details.

Box & whisker. Click the Box & whisker button to display the standard variable selection dialog box. Select variables and a box-whisker type to display a box & whisker plot based on your selections. The box and whisker plot is a useful graphical method to compare the distribution of two or more variables. Both medians and quartiles, or means and related measures of dispersion can be plotted. Note that the Box & whisker button is available only after variables have been selected.

Categorized histogram. Click the Categorized histogram button to display the standard variable selection dialog box. After you select variable(s), STATISTICA will display a categorized histogram of the distribution of the selected variable separately (categorized) for each group. Note that the Categorized histogram button is available only after variables have been selected.