Cohort
Comparison
Group A Group B (treatment)
Group A: N / events / censored
ā€“ / ā€“ / ā€“
Group A median survival
ā€“
Group B: N / events / censored
ā€“ / ā€“ / ā€“
Group B median survival
ā€“
Log-rank test (A vs B)
ā€“
How it works

Each subject gets a true event time drawn from an exponential hazard. With probability equal to the censoring slider, a subject also gets a random dropout time before the end of the study ā€” if that dropout happens first, the subject is censored: still alive when last observed, not "dead", and not thrown away.

S(t) = Ī _{t_i ā‰¤ t} (1 āˆ’ d_i / n_i)
d_i = events at time t_i, n_i = subjects still at risk at t_i
log-rank Ļ‡Ā² = (Oā‚ āˆ’ Eā‚)Ā² / V, summed per-event-time Oāˆ’E and variance
  • The step chart drops only at event times, never at censoring times.
  • Small tick marks on the curve mark exactly when a censored subject left the risk set.
  • The timeline strip above the curve shows every subject: a line ending in Ɨ (event) or | (censored).
  • In compare mode, Group B has a lower hazard (longer survival) so the curves separate ā€” the log-rank Ļ‡Ā² summarizes how unlikely that separation is under "no real difference".