2
$\begingroup$

This is a really basic question. I say Kaplan-Meier estimator and always tell people to say estimator and not method or methodology. However, I've seen a few places where statisticians also say Kaplan-Meier method/methodology. I was wondering which terminology is best and why use one over the other.

Thank you!

$\endgroup$
1
  • 2
    $\begingroup$ Method is too vague. What it is is an estimator (of the survival function). $\endgroup$
    – AdamO
    May 8 '19 at 14:50
1
$\begingroup$

I would say that you can say Kaplan-Meier estimate if you mean the statistic itself which refers to the survival function $S(t)$ (the probability that life is longer than t). But if you report a Kaplan-Meier plot (with confidence intervals), for instance, you can't call that a Kaplan-Meier estimator. A plot is no estimator, it can only be a visual representation of an estimator. I don't know whether there is an exact definition of KM method but maybe some people use the more general term KM method if they use methods associated with KM, for example if they derive the estimator and the confidence intervals and make a plot.

But I can't be sure because I don't know how people you have talked to used the term Kaplan-Meier method.

$\endgroup$
4
  • $\begingroup$ Thank you for your great reply! What if you call a plot Kaplan-Meier estimates? I would probably say KM estimates =p Here's a sentence that I usually see: "For all survival analyses, the Kaplan-Meier methodology was used, considering a 95% confidence interval (CI), aside with one-year survival rates." I would say the KM estimator was used. $\endgroup$
    – psoares
    May 8 '19 at 15:14
  • $\begingroup$ @psoares: I would say it is not correct to call a plot an estimate, because a plot is some "picture" and not any statistic like t or F value or the KM estimator. Further, the KM estimator itself is S(t) and you can calculate the confidence for the KM estimator S(t), but that doesn't mean that the KM estimator is the confidence interval. So I wouldn't call the plot nor the described methodology KM estimator. Again, I am not too sure that there is an exact definition of KM methodology/ method, so maybe people use it differently (plot with or without confidence intrvals,...) $\endgroup$
    – user247233
    May 8 '19 at 16:37
  • $\begingroup$ ... But if I think about it I see no need for such a geneal term "KM method" (if it is a used as a general term) because there is already clear definition of KM estimator and CI. So one can simply say "Figure x shows a plot with a KM estimator and 95% confidence intervals". As long as there is no clear definition of KM method I would name the things as in the sentence before $\endgroup$
    – user247233
    May 8 '19 at 17:55
  • $\begingroup$ I agree, KM method doesn't seem very specific. I prefer to use the term KM estimator. $\endgroup$
    – psoares
    May 9 '19 at 15:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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