Specifically, can I calculate a standard error from an incidence rate with known confidence intervals but no population size data?

  • $\begingroup$ Can you expand on why you think you cannot do this? Usually the confidence limits have been calculated using the standard error, possibly on a transformed scale. $\endgroup$
    – mdewey
    Mar 18 '16 at 16:55
  • $\begingroup$ I'm assuming you can, I just don't know how. The CIs have been reported in a paper, but without other info. I'm trying to work backwards to calculate the SE. Although, I think the following solves it (upper CI - Lower CI)/3.92. I thought this was only for a difference in means but it seems to work on other data that I do have numbers for. $\endgroup$
    – Incognito
    Mar 19 '16 at 19:40
  • $\begingroup$ Looks good to me. They might have used the t-test rather than the normal but that would lead to a larger number than 3.92 and hence a smaller se so you are being conservative. $\endgroup$
    – mdewey
    Mar 20 '16 at 10:18
  • 1
    $\begingroup$ Do not forget to work on the correct scale though. For instance you may have to log transform everything first. $\endgroup$
    – mdewey
    Mar 20 '16 at 10:35

Your Answer

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

Browse other questions tagged or ask your own question.