Hypothesis Testing (Non-Parametric) · TOOL GUIDE

Stratified Chi-Square Test

Open Quantly ↗

Understand the method. Use it with confidence.

Cochran-Mantel-Haenszel test controlling for a third variable. Quantly runs the method against your active analysis context and returns structured results for review and interpretation.

01

When to use it

Use this tool for rank-, count-, or distribution-based comparisons when parametric assumptions are unsuitable or variables are categorical/ordinal.
Use it only when the selected variables and study design match the method's requirements.

02

How to use it

  1. Open the tool from the Quantly Tools panel.
  2. Confirm the active dataset and analysis session.
  3. Set Categorical Variable 1 (Rows) (Required): Select the first categorical column.
  4. Set Categorical Variable 2 (Columns) (Required): Select the second categorical column.
  5. Set Stratum Variable (Optional) (Optional): If provided, and both variables are binary, performs Cochran–Mantel–Haenszel (CMH) test.
  6. Set Significance Level (α) (Required): Threshold for statistical significance.
  7. Set Generate Stratified Bar Charts (Optional): Visualize the association within each stratum using bar charts. (Used only when a stratum variable is provided).
  8. Set Generate Stacked Bar Chart (Pooled) (Optional): Visualize the overall relationship using a stacked bar chart. (Used only when no stratum is selected).
  9. Set Generate Heatmap (Pooled) (Optional): Visualize the overall contingency table as a heatmap. (Used only when no stratum is selected).
  10. Set Generate Mosaic Plot (Pooled) (Optional): Visualize the overall association using a mosaic plot. (Used only when no stratum is selected).
  11. Review optional or advanced settings and keep defaults unless your design requires a change.
  12. Run the tool, inspect warnings and diagnostics, then save or export the result with its assumptions.
03

How to interpret results

Interpret the test statistic and p-value with an effect-size measure where available. A significant result indicates evidence of a difference or association, not its practical size or causal direction.