I am new in survival analysis. I have been taught that you analyze using survival because you have a special dataset: one with censored data. And that if I work only with not censored data, I can have severe bias.

But now, when working with a dataset collected for a survival analysis, I am required to crosstab some dichotomous risk factors and death/censored data. I also have seen advice in internet to use logistic regression to predict death with the risk factors. This is supposed to be preliminary work to the real survival analysis.

My questions are: ¿Is it a good idea to run crosstabs? Why not univariate Cox if you want crude ratios?

Or exploring the data, why not Kaplan Meier instead of crosstabs?

(I am not a native english speaker. Please excuse my bad english).

Thanks in advance. Florentino Menéndez

  • $\begingroup$ You probably mean "Life tables", do you? They are proper tabulations for survival analysis. $\endgroup$
    – ttnphns
    Aug 1, 2014 at 22:59
  • $\begingroup$ Also, some applications of logistic regression are (properly) survival analysis (i.e. discrete time logit hazard models), so can you edit your question to amplify on what you mean by "use logistic regression?" $\endgroup$
    – Alexis
    Aug 2, 2014 at 0:29

1 Answer 1


Crosstabs are used for exploratory data analysis because they're quick and easy to create and have the potential to reveal a lot about the data. They can point the way to what you want to do in the "real" data analysis, i.e. Cox, Kaplan Meier, etc.


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.