Checking Proportional Hazards Assumption and Multicollinearity in Cox Proportional Hazards Model (Survival Module)
Posted: Fri Jun 20, 2025 6:38 am
Dear Jamovi Team and Community,
I am currently using the Survival module in Jamovi to perform Cox Proportional Hazards analysis. While the interface is user-friendly and effective for basic modeling, I had a couple of questions regarding model diagnostics:
Proportional Hazards Assumption:
I understand that the Cox model relies on the assumption of proportional hazards over time. Is there a way within Jamovi to check this assumption? If not, are there any recommended workarounds (e.g., exporting to R with the survival package)?
Multicollinearity Diagnostics:
In logistic regression (e.g., via the GAMLj or Regression module), we can easily check for multicollinearity using VIF (Variance Inflation Factor). However, I do not see a similar diagnostic option in the Survival module. Since multicollinearity can also affect Cox model estimates, are there any plans to implement this feature in future updates? In the meantime, would you recommend assessing multicollinearity using a separate linear model with the same predictors?
I appreciate any guidance on how best to address these two issues using Jamovi or companion tools. Thanks again for the great work on making statistical analysis more accessible.
I am currently using the Survival module in Jamovi to perform Cox Proportional Hazards analysis. While the interface is user-friendly and effective for basic modeling, I had a couple of questions regarding model diagnostics:
Proportional Hazards Assumption:
I understand that the Cox model relies on the assumption of proportional hazards over time. Is there a way within Jamovi to check this assumption? If not, are there any recommended workarounds (e.g., exporting to R with the survival package)?
Multicollinearity Diagnostics:
In logistic regression (e.g., via the GAMLj or Regression module), we can easily check for multicollinearity using VIF (Variance Inflation Factor). However, I do not see a similar diagnostic option in the Survival module. Since multicollinearity can also affect Cox model estimates, are there any plans to implement this feature in future updates? In the meantime, would you recommend assessing multicollinearity using a separate linear model with the same predictors?
I appreciate any guidance on how best to address these two issues using Jamovi or companion tools. Thanks again for the great work on making statistical analysis more accessible.