Gnumeric · Hypothesis testing · Advanced
Compare Before and After Measurements with Gnumeric
Learn when to use a paired t-test in Gnumeric and follow a five-pair example with a checkable mean difference and t statistic.
Official documentation reviewed; native application verification pending · Updated 10 October 2026 · Editorial standards
A paired t-test compares the within-pair differences between two measurements taken on the same unit. It is not the same as treating the before and after columns as two independent samples.
Practice with five matched pairs
Download matched before-and-after measurements.
| Pair | Before | After | Before − After |
|---|---|---|---|
| 1 | 12 | 10 | 2 |
| 2 | 15 | 13 | 2 |
| 3 | 14 | 11 | 3 |
| 4 | 18 | 16 | 2 |
| 5 | 16 | 13 | 3 |
The difference vector is [2,2,3,2,3], with mean 2.4, sample standard deviation √0.3 ≈ 0.5477, and standard error √0.3/√5 ≈ 0.24495. Therefore t ≈ 9.798 on 4 degrees of freedom, when differences are defined as Before minus After.
Use a paired-sample analysis
- Import the data, with the paired measurements aligned by row; do not sort one of the two measurement columns separately.
- From Gnumeric’s Statistics menu, look for a paired two-sample t-test or similarly named analysis tool, as covered by its official statistical-analysis manual.
- Select the matching Before and After data ranges. Include headers only if you enable labels.
- Set a hypothesized mean paired difference of zero for this teaching exercise.
- Inspect the reported mean difference, t statistic, degrees of freedom and the chosen one-sided/two-sided significance result.
Important sign convention
If you reverse the column order, the mean difference and t statistic change sign. That is not a numerical error: After − Before has mean −2.4 and t approximately −9.798. The magnitude of the two-sided t statistic stays the same.
Interpretation and common errors
The test assumes differences are meaningful matched observations and that model assumptions are reasonable. With only five pairs, you should inspect the raw differences, not rely on the p-value alone. Treat this as a calculation tutorial, not evidence for a clinical, scientific, or business intervention.
Related: Gnumeric one-way ANOVA compares independent groups rather than matched pairs.
Verification: Paired differences, standard deviation, degrees of freedom and t were independently checked. Exact Gnumeric dialog labels remain native-app test items.
Official documentation: Software manual or vendor release notes. The workflow follows the linked documentation; confirm the interface and outputs in your installed version because no native-application test is documented for this guide.