Which test is used for 3 or more related samples in nominal data?

The appropriate test for analyzing three or more related samples in nominal data is the Friedman two-way ANOVA. This test is designed to handle repeated measures when the samples are related and helps to assess whether there are statistically significant differences in the median of three or more related groups. It is particularly useful when the data violate the assumptions of normality required for a traditional ANOVA, making it a non-parametric alternative.

In the context of nominal data, the Friedman test evaluates the ranks of the data across related groups to determine if at least one group is statistically different from the others. This ability to assess rank differences across multiple related samples is essential in research scenarios where repeated measures are involved, such as patient responses over different time points or conditions.

The other options listed do not appropriately address the requirement for three or more related samples in nominal data. The McNemar test is suitable for two related samples, while the Wilcoxon Signed Rank test is used for two related samples as well, specifically for ordinal data or continuous data that is not normally distributed. The Chi-square test typically assesses relationships between categorical variables but does not handle repeated measures or related samples, specifically when nominal data is involved across three or more groups.