Statistical Test Selection: Decision Tree
Quick Reference
Follow the flowchart below by answering simple questions about your data. Each path leads you to the appropriate statistical test.
Decision Flowchart
graph TD
A[What type of OUTCOME variable?] --> B[Continuous]
A --> C[Categorical]
B --> D[How many groups?]
D --> E[1 group]
D --> F[2 groups]
D --> G[3+ groups]
D --> H[Continuous predictor]
E --> I[One-Sample t-test]
F --> J[Same people twice?]
J --> K[Yes: Paired t-test]
J --> L[No: Independent t-test]
G --> M[Same people all groups?]
M --> N[Yes: Repeated Measures ANOVA]
M --> O[No: One-Way ANOVA]
H --> P[Regression]
C --> Q[Chi-Square Tests]
Text-Based Decision Tree
START: What type of OUTCOME variable do you have?
→ CONTINUOUS (numbers with meaning)
How many GROUPS do you want to compare?
Compare to known value?
→ YES: One-Sample t-test
Compare your sample mean to a known population value (e.g., national average)
Same people measured twice?
→ YES: Paired t-test
Same participants at two time points or conditions
→ NO: Independent t-test
Different people in each group
Assumptions violated?
Use Mann-Whitney U test instead
Same people across all groups?
→ YES: Repeated Measures ANOVA
Assumptions violated?
Use Friedman Test instead
→ NO: One-Way ANOVA
Assumptions violated?
Use Kruskal-Wallis test instead
Two grouping variables?
→ Two-Way ANOVA
- Test interaction? Use
* - Main effects only? Use
+
Want to predict outcome from predictor(s)?
→ ONE predictor: Simple Linear Regression
→ MULTIPLE predictors: Multiple Regression
Assumptions violated?
Consider transformation or robust methods
→ CATEGORICAL (groups/categories)
How many variables?
Chi-Square Goodness-of-Fit
Compare observed frequencies to expected frequencies
Chi-Square Test of Independence
Are two categorical variables related?
Quick Reference Table
| Your Situation | Test to Use | Section | Key Assumptions |
|---|---|---|---|
| Compare 1 group to known value | One-Sample t-test | T-Tests | Normality |
| Compare 2 independent groups | Independent t-test | T-Tests | Normality, independence |
| Compare same people twice | Paired t-test | T-Tests | Normality of differences |
| Compare 3+ independent groups | One-Way ANOVA | ANOVA | Normality, equal variance |
| Compare groups with 2 factors | Two-Way ANOVA | ANOVA | Normality, equal variance |
| Same people, 3+ times | Repeated Measures ANOVA | ANOVA | Sphericity, normality |
| Predict continuous from continuous | Regression | Regression | Linearity, normality of residuals |
| Relate 2 continuous variables | Pearson Correlation | Regression | Linearity, bivariate normality |
| Compare frequencies/counts | Chi-Square | Chi-Square | Expected counts ≥ 5 |
| Non-normal, 2 groups | Mann-Whitney U | Non-Parametric | Independence |
| Non-normal, 3+ groups | Kruskal-Wallis | Non-Parametric | Independence |
| Non-normal, paired data | Wilcoxon Signed-Rank | Non-Parametric | Symmetric differences |
Step-by-Step Questions
Question 1: What is my OUTCOME variable?
- Continuous (measurements, scores, times) → Go to Question 2
- Categorical (groups, yes/no, categories) → Chi-Square
Question 2: What is my PREDICTOR variable?
- Groups/Categories → Go to Question 3
- Continuous (another measurement) → Regression/Correlation
Question 3: How many groups?
- 1 group (comparing to known value) → One-Sample t-test
- 2 groups → Go to Question 4
- 3+ groups → Go to Question 5
Question 4: Are the two groups independent?
- Same people measured twice → Paired t-test
- Different people in each group → Independent t-test
- Check normality → If violated, use Mann-Whitney U
Question 5: Are there 3+ groups?
- Same people in all groups → Repeated Measures ANOVA
- Different people in each group → One-Way ANOVA
- Two factors (2 grouping variables) → Two-Way ANOVA
- Check normality → If violated, use Kruskal-Wallis
Still Unsure?
Check out Common Scenarios for real research examples!