Researchers interested in the effectiveness of a new tumor-suppression medication recorded time to tumor for cancer patients. Forty-three patients were treated with an older drug (A) and 76 were treated with the new drug (B).
The survival functions cross.
LogRankTest does not detect a significant difference in the underlying hazard rates.
Suppose researchers expected that the new drug would be more effective at preventing late-developing tumors. The following Fleming-Harrington weights emphasize late survival times.
LogRankTest weighted for late survival detects a significant difference.