Multicollinearity & VIF
When X variables are highly correlated, the model cannot separate their individual effects — standard errors inflate and coefficients become unstable. VIF (Variance Inflation Factor) quantifies this for each predictor:
- VIF < 5: clean — no multicollinearity concern
- VIF 5–10: moderate — review variable selection
- VIF > 10: severe — remove or combine the variable
- Remove one of the correlated variables
- Combine them (ratio, difference, index)
- Collect more diverse data
Try it in the app
Try: Run regression with 2+ X variables → check the VIF column in Results → Coefficients table. Values >5 deserve attention.