Schoenfeld residuals can be used to test the proportionality of the model as a whole. There is also an option to test the proportionality of each x-variable.
More informationhelp |
Practical example
The first step is re-run the multiple Cox regression model. The quietly option is included in the beginning of the command to suppress the output.
quietly stcox gpa sex ib1.marstat40 if pop_cox==1, noshow |
Next, we try the estat phtest command.
estat phtest, detail |
If the p-value (Prob>chi2) is below 0.05, it means that we should reject the proportionality assumption. The global test suggests this model violates the proportionality assumption. As indicated by the p-values for the x-variables, our problem seems to be marstat40. We will not explore this further, but a solution might be to transform marstat40 into a binary variable instead (e.g. Married vs. not Married).