Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • 95% confidence interval for difference in two estimates

    Hi,

    I'm reading a paper that calculated the difference in two percentages with 95%CI using Stata 14. I find this very interesting, but not finding any example to do this by myself. Can anyone please provide an example or the command to do this.

    Thanks in advance!

    Click image for larger version

Name:	Capture d’écran 2020-09-13 à 11.21.00.png
Views:	1
Size:	146.4 KB
ID:	1572633

  • #2
    There is -prtest- which relies on large-sample asymptotics. Then you could implement a similar estimate and test in a regression framework. Then the authors may have programmed their own variance estimate.

    There are many ways to compute the difference between two proportions, some use variance stabilizing transformations, others rely on transformations to the risk or odds ratio scale and back transform. Confidence intervals can also be derived by inversion of the Score test.

    Comment


    • #3
      Answer retracted, Leonardo's is better.

      Comment


      • #4
        Originally posted by Leonardo Guizzetti View Post
        There is -prtest- which relies on large-sample asymptotics. Then you could implement a similar estimate and test in a regression framework. Then the authors may have programmed their own variance estimate.

        There are many ways to compute the difference between two proportions, some use variance stabilizing transformations, others rely on transformations to the risk or odds ratio scale and back transform. Confidence intervals can also be derived by inversion of the Score test.
        Thanks indeed Leonardo for the description!

        Comment

        Working...
        X