I have run a cox regression with only a treatment in mind so far and have then tested the proportional hazards assumption along with a graphical assessment using the log-log survival plot. My global test seems to be fine (p-value = 0.6918) but the graph has my two lines crossing each other quite a bit.
What do you do in this scenario? I was under the impression that the two lines shouldn't cross so I am a bit confused as to why the global test would be OK then.
Any help would be greatly appreciated!