Wilcoxon signed-rank, sign and McNemar tests
Compares two related measurements (before/after, matched pairs). Wilcoxon uses the size and sign of the differences, the sign test only their sign, McNemar compares paired yes/no outcomes.
Menu: Analyze → Nonparametric Tests → Legacy Dialogs → 2 Related Samples…
When should I use it?
- Pre/post scores that are not normally distributed.
- Did the share passing change between two attempts (McNemar)?
What are the assumptions?
- Paired observations.
- Wilcoxon: differences roughly symmetric.
How do I run it in StatLab?
- Put the first variables of each pair in 'Variable 1' and the second ones, in the same order, in 'Variable 2'.
- Tick the tests wanted.
How do I read the output?
| Output | Meaning |
|---|---|
| Negative / Positive Ranks / Ties | Cases where the second value is lower / higher / equal. |
| Z, Asymp. Sig. | Based on the smaller rank sum (footnote). |
| McNemar Exact Sig. | Binomial test on the discordant pairs. |
Worked example
On the practice survey dataset (File → Sample dataset), this command — which the dialog's Paste button also produces — runs the procedure:
PROC np2Related /first=(score age) /second=(pre post) /wilcoxon=YES /sign=NO /mcnemar=NO /exact=NO.
Output tables: Ranks · Test Statistics.
How would I report it in APA style?
A Wilcoxon signed-rank test for Pre-test score - Wellbeing score gave Z = −10.48, p < .001, r = .86. A Wilcoxon signed-rank test for Post-test score - Age (years) gave Z = −7.08, p < .001, r = .58.
Computed by StatLab's validated procedure code when this page was built. The example only illustrates the steps; choose variables that fit your own research question.
What are common mistakes?
- Using these tests for independent groups.
- Forgetting that ties (zero differences) are dropped from Wilcoxon and sign tests.