Announcement

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

  • how to export logit confidence interval after proportions estimation?

    Dear fellows,
    I want to export the logit confidence interval that is estimated by the proportion command. But using estout, I only obtain a confidence interval that is based on a normal distribution, not the interval that is presented in the results window.
    I would be very glad about suggestions how I could export the correctly estimated confidence interval.
    Thanks a lot!

    I promted
    svy, subpop(if l11101==9 & HVP==1 ): proportion s_wirtsituation if syear==2022
    eststo M1
    estout M1 , cells(ci(fmt(3)))


    resulting in
    Survey: Proportion estimation

    Number of strata = 102 Number of obs = 16,862
    Number of PSUs = 4,486 Population size = 64,459,723
    Subpop. no. obs = 2,201
    Subpop. size = 9,132,892.2
    Design df = 4,384

    ----------------------------------------------------------------------
    | Linearized Logit
    | Proportion std. err. [95% conf. interval]
    ---------------------+------------------------------------------------
    s_wirtsituation |
    keine großen Sorgen | .8560271 .0146033 .8249679 .8823608
    große Sorgen | .1439729 .0146033 .1176392 .1750321
    ----------------------------------------------------------------------
    Note: 123 strata omitted because they contain no subpopulation
    members.
    Note: Strata with single sampling unit centered at overall mean.

    . eststo M1

    . estout M1 , cells(ci(fmt(3)))

    -------------------------
    M1
    ci95
    -------------------------
    0.s_wirtsi~n 0.827,0.885
    1.s_wirtsi~n 0.115,0.173
    -------------------------






  • #2
    Dear fellows,

    I want to draw a graph on estimated parameters with confidence intervals. Since I have annual survey data that needs to be weighted, I work in the svy environement.
    To obtain the parameters, which I want to plot, I want to invoke the proportions command in loops over several variables for several subgroups over several years and write the results in an txt-file that I import in a next step into Stata and then graph the estimated parameters with the twoway line and twoway rarea commands.
    This works very well with means and confidence intervals around them.
    But when I estimate proportions (see code below), estout exports the wrong confidence intervals; the exact same is happens with esttab.
    In the results window, Stata presents the default logit confidence interval together with the estimated proportion, but using estout or esttab, I obtain a confidence interval that is based on a normal distribution, which is neither correct nor presented in the results window.

    I would be very glad about suggestions how I could export the estimated and presented logit confidence interval with the estout command (estout is a user wirtten command)
    Thanks a lot!

    I promted
    Code:
    svy, subpop(if var1==1 & var2==1 ): proportion var3 if syear==2022
    eststo M1
    estout M1 , cells(ci(fmt(3)))
    resulting in
    Code:
    ----------------------------------------------------------------------
                         |                Linearized              Logit
                         | Proportion     std. err.        [95% conf. interval]
    ---------------------+------------------------------------------------------
    var3                 |
    zero                 | .8560271       .0146033         .8249679 .8823608
    one                  | .1439729       .0146033         .1176392 .1750321
    ----------------------------------------------------------------------------
    
    [...]
    
    . eststo M1
    
    . estout M1 , cells(ci(fmt(3)))
    
    -------------------------
                          M1
                        ci95
    -------------------------
    0.var3       0.827,0.885
    1.var3       0.115,0.173
    -------------------------
    Last edited by Susanne Elisabeth Elsas; Today, 03:21.

    Comment

    Working...
    X