Non-Parametric Tests in Excel
Use non-parametric tests when data is:
- Counts or frequencies of different types
- Measured on nominal or ordinal scale
- Not meeting assumptions of a normal test
- Distribution is unknown
- A small sample
- Imprecise
- Skewed data that make the median more representative
Note: Excel doesn't have the ability to do statistical tests of non-normal (i.e., not "bell shaped") data. QI Macros, however, have a set of templates to handle non-parametric data.
| Samples | |||
| Variable (Continuous) Data | 1 | 2 | 3 |
| Independent | Wilcoxon | ||
| Dependent | Spearman | ||
| Goodness of Fit | Kolmogorov-Smirnov (K-S) | Tukey | |
| Attribute (Ordinal Data | |||
| Independent | 1-Sample Sign Test | Sign Test | |
| Dependent | Spearman | ||
Non-Normal vs Normal Tests |
|
| Non-parametric | Parametric (Normal) |
| 1-sample sign | 1-sample z-test 1-sample t-test |
| Sign Test | Paired two-sample t Test |
| Wilcoxon Signed Rank Test | |
| Mann-Whitney Wilcoxon Rank Sum Test |
2-sample t-test |
| Kruskal-Wallis | one-way ANOVA |
| Mood's Median | one-way ANOVA |
| Friedman | 2-way ANOVA test |
