Framework
Choosing a statistical test
The question it answers
When analysing survey or trial data, which statistical test should an analyst choose to fit the question being asked and the type of data at hand?
- Origin
- Kim, Kim and Kwak, 2024
- Diagram type
- Comparison table
- Level of analysis
- Individual
- Topics
- Data & analytics
Diagram
Choosing a statistical test
| Criterion | Question | Variable type | Usual test |
|---|---|---|---|
| Two independent groups | Do the means differ? | Quantitative outcome, categorical group | Independent t-test (or non-parametric if not normal) |
| Before and after | Did the same units change after an intervention? | Two paired measurements | Paired t-test |
| Three or more groups | Does any group differ? | Quantitative outcome, categorical group | ANOVA (or Kruskal–Wallis) |
| Two categorical variables | Are they associated? | Two nominal variables | Chi-square |
| Two quantitative variables | How do they move together? | Two numeric variables | Pearson correlation, simple regression |
| Yes/no outcome | Which factors predict it? | Binary outcome | Logistic regression |
Pitfalls
Pitfall
Using a parametric test when assumptions fail; ignoring the measurement scale.
Sources
References
Listed A–Z by author surname, Harvard style. A source with a DOI links to https://doi.org/…; one without links to its page or says plainly that it has no DOI.
- Kim, J., Kim, D.H. and Kwak, S.G. (2024) ‘Comprehensive guidelines for appropriate statistical analysis methods in research’, Korean Journal of Anesthesiology, 77(5), pp. 503–517.doi:10.4097/kja.24016Source page