Announcement

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

  • how to obtain 95%CI for difference in proportions from logistic regression output

    Hi,

    Please can you help me to calculate the 95%CI for the difference in proportions from logistic regression?

    This is the example

    webuse mheart1s20
    mi describe

    prtest attack, by(female)
    logit attack i.female, or
    margins i.female, pwcompare

    mi estimate, or: logit attack i.female


    As you can see using margins after logistic I can obtain the difference and confidence interval in line with the prtest (difference in proportions)

    I know how to calculate the proportions of attack in males and females from odds and odds ratios of the output. However, I don't know how to use solely the logit output to obtain the confidence intervals for the difference in proportions.

    I am using multiple imputation and multiple imputation does not allow me to use margins, so I cannot predict the confidence interval for the difference in proportions.

    So, how can I use the output of logistic to obtain the confidence intervals in the difference of proportions scale for the following output


    mi estimate, or: logit attack i.female
    ------------------------------------------------------------------------------
    attack | Odds Ratio Std. Err. t P>|t| [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    1.female | .996337 .3745198 -0.01 0.992 .4769165 2.08147
    _cons | .8125 .1516913 -1.11 0.266 .5635164 1.171494
    ------------------------------------------------------------------------------

    Thank you in advance for any help.

    Andrew
Working...
X