Rank-hazard plot interpretation

f2harrell: > smoothed scaled Shoenfeld residuals plots Thanks, but I’m still don’t understand how this relates to the non-equally spaced y-axis in the rank hazard plot(hazard ratio vs rank), and how that scaling give “meaningful hazard ratio plots vs time”. If there are specific references (e.g., sections of the R survival package vignettes) that explain this connection, I’d be grateful if you could point me to them.