Skip to contents

Use this flow for the main scope of version 1.0.0: group comparison.

flowchart TD
    A["Start"] --> B{"What do you want?"}
    B -->|Compare groups| C{"Outcome type"}
    B -->|Anything else| Z["Redirect to another analysis branch"]

    C --> C1["Continuous"]
    C --> C2["Categorical"]
    C --> C3["Ordinal"]
    C --> C4["Count"]

    C1 --> D{"How many groups?"}
    D -->|2| E{"Paired?"}
    D -->|3+| F{"Repeated?"}
    E -->|No| G{"Normal enough?"}
    E -->|Yes| H{"Normal enough?"}
    G -->|Yes| T1["Welch t-test"]
    G -->|No| T2["Mann-Whitney U test"]
    H -->|Yes| T3["Paired t-test"]
    H -->|No| T4["Wilcoxon signed-rank test"]
    F -->|No| I{"Normal enough?"}
    F -->|Yes| J{"Normal enough?"}
    I -->|Yes| T5["Welch ANOVA"]
    I -->|No| T6["Kruskal-Wallis test"]
    J -->|Yes| T7["Repeated-measures ANOVA"]
    J -->|No| T8["Friedman test"]

    C2 --> K{"Paired?"}
    K -->|No| L{"Expected counts small?"}
    K -->|Yes| T9["McNemar test"]
    L -->|Yes| T10["Fisher's exact test"]
    L -->|No| T11["Chi-squared test"]

    C3 --> M{"Paired?"}
    M -->|No, 2 groups| T12["Mann-Whitney U test"]
    M -->|Yes, 2 groups| T13["Wilcoxon signed-rank test"]
    M -->|No, 3+ groups| T14["Kruskal-Wallis test"]
    M -->|Yes, 3+ groups| T15["Friedman test"]

    C4 --> T16["Stop simple testing and move to count regression"]

Practical rule

  • If you only want a standard group comparison, stay on this page.
  • If you need adjustment, survival analysis, agreement, or equivalence, statsguider should stop and redirect you.