How to Choose a Statistical Test

A practical decision tree to help you select the right statistical test for your research question and data type.

Interactive Decision Tree

Step 1: What is your research question?

Testing relationships between variables

Example: "Is there a relationship between study hours and exam scores?"

Comparing means/medians across groups

Example: "Do treatment and control groups differ in outcomes?"

Predicting an outcome from predictors

Example: "Can we predict GPA from study time and motivation?"

Quick Reference Table

Research Question	Data Type	Number of Groups	Recommended Test
Relationship	Both continuous	—	Pearson Correlation or Linear Regression
Relationship	Both categorical	—	Chi-Square Test
Group comparison	Continuous outcome	2 independent	Independent t-test
Group comparison	Continuous outcome	2 paired	Paired t-test
Group comparison	Continuous outcome	3+ independent	One-Way ANOVA
Prediction	Continuous outcome	—	Linear Regression

Common Pitfalls to Avoid

❌ Running multiple t-tests instead of ANOVA

When comparing 3+ groups, use ANOVA instead of multiple t-tests. Multiple comparisons inflate Type I error rate.

❌ Using parametric tests with non-normal data

Check your data distribution. For highly skewed data or small samples, consider non-parametric alternatives (Mann-Whitney U, Kruskal-Wallis, Spearman correlation).

❌ Confusing correlation with causation

Correlation and regression show relationships, not causation. Only randomized controlled trials can establish causality.

❌ Treating paired data as independent

Before/after measurements or matched pairs require paired tests. Using independent tests loses statistical power and may give incorrect results.

Need more statistical tools?

Explore Professional Analysis Tools →