I have 45 coral colonies undergoing three different treatments (n = 15) and surveyed their average survival after 1,2 and 3 months. I believe I cannot use Kaplan-Meier, as I do not know the exact time of death. In addition, mortality was low and usually only partially (e.g. 20% of a colony). The survival data look something like: Colony1: 100% (month0) - 99% (month1) - 97% (month2) - 97% (month3); Colony2: 100% (month0) - 96% (month1) - 96% (month2) - 94% (month3); ...etc

Is there any survival analysis I can apply to such data, to distinguish survival between the three treatments. Or should I just try to analyse it with a repeated ANOVA?


1 Answer 1


For Kaplan Meier you don't need a date like 3rd of October 2016, it would be fine enough to have consecutive numbers, which indicate your survey time structure. Would like to comment, but I am too new here. Furthermore, If you have percentage data of your group, and they are only 15 people, it is fairly easy to calculate how many people actually survived each treatment, and you can use this figures instead.

However: If you have covariates, which do differ on the individual level and you only have group percentages, my suggestion here obviously doesn't work.

  • 1
    $\begingroup$ Discrete time survival models may help. You may have interval censoring as it seems that what you have is "current status" data. $\endgroup$ Oct 3, 2016 at 11:33
  • $\begingroup$ Concerning your beamer problem: please try pastebin.com/5CgzeHKp $\endgroup$
    – user248993
    May 24, 2019 at 19:55
  • $\begingroup$ @samcarter thank you very much. However, it does not work for me. I just ran your code and I still have the number 1 on both slides. I probably stick with that... $\endgroup$ May 29, 2019 at 16:37
  • $\begingroup$ @user82729 I just tried a few things and indeed this does not work with old beamer versions. You need beamer version 3.54 or newer $\endgroup$
    – user248993
    May 29, 2019 at 20:40

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.