I am building a Cox Proportional Hazards Model to predict the survival outcome of seabird faced with predation pressure. I have 6 factor variables with two or three levels each that I have predicted to affect survival. Three of which are management relevant (they can be manipulated by wildlife managers to increase or decrease survival if significant). The ultimate goal of the model is prediction but I would like to include the management relevant variables as well even if not significant. How can I check for multicollinearity among my variables. I am using program R for the analysis.

**Contents**hide

#### Best Answer

(Almost) perfect multicollinearity (MC) will lead to large standard errors of the estimates and/or non-convergence of the optimization routines. Any other MC is no issue since your goal is prediction (and not interpretation of effects).

### Similar Posts:

- Solved – Diagnosing collinearity in a Cox proportional hazards model
- Solved – Diagnosing collinearity in a Cox proportional hazards model
- Solved – Testing proportional hazards assumption in parametric models
- Solved – Testing proportional hazards assumption in parametric models
- Solved – Crossing survival curves hypothesis test