What is the earliest paper that proposed the stratified version of the log-rank test?

Unfortunately, finding citations/references in textbooks for established statistics procedures is difficult.


1 Answer 1


As mentioned by Jun Shao in his paper, it was by R Peto et al. in their paper Design and analysis of randomized clinical trials requiring prolonged observation of each patient. I. Introduction and design, the description being found in PubMed site.


In fact, Peto & Peto in an earlier paper Asymptotically Efficient Rank Invariant Test Procedures, used the term log-rank for their invariant rank based statistic. Later, they generalized for "multi-group experiments".

  • $\begingroup$ I have already reviewed this reference, and I do not see any section discussing the development of the statistic in question. $\endgroup$ Commented Apr 11 at 4:53
  • $\begingroup$ Check their earlier paper linked above. $\endgroup$ Commented Apr 11 at 10:23
  • 1
    $\begingroup$ No, it's much earlier by Nathan-Mantel under the Mantel-Haenszel framework. $\endgroup$ Commented Apr 11 at 12:14
  • 2
    $\begingroup$ Yes, @FrankHarrell, you are right in that Peto wasn't the first in developing the statistic based on chi square (even Peto cited their 1959 paper) . But it was Peto's 1972 paper, which according to Kalbfleisch and Prentice, was the first to "coin the term log-rank". $\endgroup$ Commented Apr 11 at 12:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.