Exporting really simple statistics (ci and proportions) of a dummy, by categorical variable

[Nanna Lauritz]

Dear Statalists,

I'm trying to export a table of information from Stata but I cannot find a solution. I have a variable list of 20 variables, all of them are dummies. For example: education is equal to 1 if the respondent mentions education and 0 otherwise. I have 19 other similar variables.
I want to see the propensity for respondents preferring Parties A, B, C, D or E (categorical) to mention variable 1(..., N) on my variable list.

I can easily do this in stata:

by party: ci prop $dictionary

But I can not find out how to export this.
I have tried asdoc (which I think isn't working because I have too many variables in my $dictionary), esttout, outreg2 and dtable. None of it seems to be capable of giving me the output that I want. Do any of you have an idea for a solution?

[Henry Strawforrd]

How about this?
http://repec.org/bocode/e/estout/estpost.html

[Nanna Lauritz]

How about this?
http://repec.org/bocode/e/estout/estpost.html

 by party: eststo: ci prop $dictionary2

Thanks. At first glance this would seem to work - as it does still do the 'by party: ci prop $dictionary2", but I can't specify adding percentages (prop) to the output. Only CI.
Also whatever it is that Stata stores in est* definitely isn't what went into the eststo command.

[Jeff Pitblado (StataCorp)]

Here is an example of how this might be done with collect. The
dummy variables in this example have missing values, so if that is true
in your data, you might want to add the sample sizes to the table, but I
will leave that up to you.

webuse nhanes2l, clear
sort region

collect style use dtable, replace

global vlist heartatk diabetes highlead highbp
foreach var of global vlist {
    collect, tags(var[`var']) : by region : ci prop `var'
}

* show list of collected results
collect levelsof result

* build and style composite result for the CIs
collect composite define ci = lb ub, notrim
collect style cell result[ci], sformat("[%s]")
collect style cell result[proportion ci], nformat("%5.3f")

* result auto levels for the layout
collect style autolevels result proportion ci

* use shorter labels
collect label levels result proportion "Prop" ci "%95 CI", modify

collect style column, dups(center)

collect layout (var) (region#result)

Here is the resulting table.

. collect layout (var) (region#result)

Collection: default
      Rows: var
   Columns: region#result
   Table 1: 4 x 8

---------------------------------------------------------------------------------------------------
                             NE                  MW                  S                   W
                     Prop    [%95 CI]    Prop    [%95 CI]    Prop    [%95 CI]    Prop    [%95 CI]
---------------------------------------------------------------------------------------------------
Prior heart attack  0.037 [0.029 0.046] 0.044 [0.036 0.052] 0.046 [0.039 0.054] 0.056 [0.047 0.065]
Diabetes status     0.047 [0.038 0.057] 0.045 [0.038 0.053] 0.056 [0.048 0.066] 0.044 [0.036 0.052]
High lead level     0.067 [0.052 0.084] 0.071 [0.057 0.086] 0.045 [0.035 0.058] 0.056 [0.044 0.070]
High blood pressure 0.441 [0.419 0.462] 0.400 [0.382 0.419] 0.427 [0.408 0.445] 0.428 [0.409 0.447]
---------------------------------------------------------------------------------------------------

[Nanna Lauritz]

Here is an example of how this might be done with collect. The
dummy variables in this example have missing values, so if that is true
in your data, you might want to add the sample sizes to the table, but I
will leave that up to you.

webuse nhanes2l, clear
sort region

collect style use dtable, replace

global vlist heartatk diabetes highlead highbp
foreach var of global vlist {
collect, tags(var[`var']) : by region : ci prop `var'
}

* show list of collected results
collect levelsof result

* build and style composite result for the CIs
collect composite define ci = lb ub, notrim
collect style cell result[ci], sformat("[%s]")
collect style cell result[proportion ci], nformat("%5.3f")

* result auto levels for the layout
collect style autolevels result proportion ci

* use shorter labels
collect label levels result proportion "Prop" ci "%95 CI", modify

collect style column, dups(center)

collect layout (var) (region#result)

Here is the resulting table.

. collect layout (var) (region#result)

Collection: default
Rows: var
Columns: region#result
Table 1: 4 x 8

---------------------------------------------------------------------------------------------------
NE MW S W
Prop [%95 CI] Prop [%95 CI] Prop [%95 CI] Prop [%95 CI]
---------------------------------------------------------------------------------------------------
Prior heart attack 0.037 [0.029 0.046] 0.044 [0.036 0.052] 0.046 [0.039 0.054] 0.056 [0.047 0.065]
Diabetes status 0.047 [0.038 0.057] 0.045 [0.038 0.053] 0.056 [0.048 0.066] 0.044 [0.036 0.052]
High lead level 0.067 [0.052 0.084] 0.071 [0.057 0.086] 0.045 [0.035 0.058] 0.056 [0.044 0.070]
High blood pressure 0.441 [0.419 0.462] 0.400 [0.382 0.419] 0.427 [0.408 0.445] 0.428 [0.409 0.447]
---------------------------------------------------------------------------------------------------

Thanks Jeff. Your code works perfectly, but I wanted to have this kind of table in my paper. How do I export this kind of table?

[Rich Goldstein]

help collect export

[Nanna Lauritz]

Yep sorry should have done that.
Here's the code for future readers:

collect export party.docx, as(docx)