1 · Upload your metadata
Start with your sample sheet. Omics measurements are not .
Screening thresholds
| Q1 comparison | Association measure | Priority |
|---|---|---|
| Categorical–categorical | Cramér’s V | ≥ 0.40 |
| Categorical–continuous | Eta | ≥ 0.40 |
| Continuous–continuous | Absolute Spearman correlation | ≥ 0.40 |
If no variable meets the main threshold, associations from 0.20 to below 0.40 are shown. These are teaching priorities, not proof of confounding. SMD and Duncan do not select variables at Q1.
After Q2/Q3 flag a potential confounder, examine shared categories and distributions/ranges. SMD and Duncan support imbalance assessment. High imbalance or correlation alone does not establish complete confounding.
Q1 is descriptive: related specimens remain in the screen, so biological units contributing more specimens may influence it more. Technical repeats are counted once using specimen ID where available. When “Same specimen” is selected without specimen IDs, the biological-unit ID is used.
No sheet loaded.
2 · Map the key columns
3 · Choose variables to screen
Choose metadata variables and check their types. Identifiers and the outcome are not screened as candidate variables.