Announcement

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

  • how does stata calculate confidence intervals for age standardised means/proportions

    Dear Statalisters,

    I'm using the following Stata command (see help mean):
    mean x, std(a) stdw(b) over(c)

    to get age-standardised mean or proportion of a variable of interest x. a= variable for age-group and b= variable for weights for each group. x can be a continuous variable or a binomial variable. My question is how does Stata calculate 95% CI for such direct age-standardised means/proportions? What method/approximation is used? I tried to look up relevant help files under Methods but couldnt really understand. Please help. Many

    Thanks Amit A, UNSW.,

  • #2
    Did you look at "Methods and Formulas" in the Stata online manual? You can get to that from the help file by clicking on the link in the upper left corner of the help window. The formula you need is on page 1273 of the manual.
    Richard T. Campbell
    Emeritus Professor of Biostatistics and Sociology
    University of Illinois at Chicago

    Comment


    • #3
      Thanks Richard
      Yes I did look it up but unfortunately I cant really understand that formula (being untrained in advanced Maths!)
      What is it basically doing and how is it different to ordinary method of calculating 95% ci? Thanks again.

      Comment


      • #4
        I think you can see how it is different if you were to assume that each case had a weight of 1. If that were the case, the sum of the weights would be N. If you plug that into the formulas,that is, substitute N for W where ever it appears and 1 for wi you should see that you have conventional formulas for the mean, standard error of the mean and CI's. When the weights are not all equal to 1 some cases get more influence than others, but you should be able to see that the concepts are the same as you learned in your introductory statistics course..
        Richard T. Campbell
        Emeritus Professor of Biostatistics and Sociology
        University of Illinois at Chicago

        Comment


        • #5
          OK great, many thanks

          Comment

          Working...
          X