dtable Stata18 - Row percentages

[Elena Mylona]

Hello!

I am using dtable which provides column percentages, however, I also want to get row percentages. Do you know any way I can do that without doing, for example, this, dtable i.nonresponse, by(sex), for all my variables separately and then filling my table manually? I have 10 variables in total, separated by nonresponse (binary variable).

Thank you!

[Jeff Pitblado (StataCorp)]

I'll use the NHANES II data to show how to use table to get
tables similar to dtable but with row percentages across a common
by variable instead of column percentages within levels of the
by variable.

In the following I'm using rural as the binary by variable.
The other variables in local macro vlist are the factor variable you
would normally use with dtable to get column percentages.

webuse nhanes2l

unab vlist : smsa sex race hlthstat heartatk diabetes highlead

* build the overall sample table
table () (rural), ///
    stat(frequency) ///
    stat(percent, across(rural)) ///
    name(N)
collect addtags N[_hide]
collect layout (N) (rural#result)

* loop over the factor variables to compute their "row" percentages
* across the levels of -rural-
foreach v of local vlist {
    table (`v') (rural), ///
        stat(frequency) ///
        stat(percent, across(rural)) ///
        totals(`v') ///
        name(`v')
}

* create a new collection named 'all' that combines the above collections
collect combine all = N `vlist'

* define a composite result like -dtable- does
collect composite define stats = frequency percent

* add formats for the percent like -dtable- does
collect style cell result[percent], nformat("%6.2f") sformat("(%s%%)")

* hide the title and level labels in the header for our composite result
collect style header result[stats], title(hide) level(hide)

* new layout putting all the factor variables in the rows
collect layout (N `vlist') (rural#result[stats])

Here is the resulting table.

. collect layout (N `vlist') (rural#result[stats])

Collection: all
      Rows: N smsa sex race hlthstat heartatk diabetes highlead
   Columns: rural#result[stats]
   Table 1: 27 x 3

------------------------------------------------------------------------------
                         |                         Rural                      
                         |           Urban            Rural              Total
-------------------------+----------------------------------------------------
N                        |  6,548 (63.26%)   3,803 (36.74%)   10,351 (100.00%)
SMSA type                |                                                    
  SMSA, central city     |  2,629 (99.89%)        3 (0.11%)    2,632 (100.00%)
  SMSA, not central city |  2,360 (78.46%)     648 (21.54%)    3,008 (100.00%)
  Not in an SMSA         |  1,559 (33.09%)   3,152 (66.91%)    4,711 (100.00%)
Sex                      |                                                    
  Male                   |  3,023 (61.51%)   1,892 (38.49%)    4,915 (100.00%)
  Female                 |  3,525 (64.85%)   1,911 (35.15%)    5,436 (100.00%)
Race                     |                                                    
  White                  |  5,419 (59.78%)   3,646 (40.22%)    9,065 (100.00%)
  Black                  |    968 (89.13%)     118 (10.87%)    1,086 (100.00%)
  Other                  |    161 (80.50%)      39 (19.50%)      200 (100.00%)
Health status            |                                                    
  Excellent              |  1,609 (66.85%)     798 (33.15%)    2,407 (100.00%)
  Very good              |  1,713 (66.11%)     878 (33.89%)    2,591 (100.00%)
  Good                   |  1,878 (63.92%)   1,060 (36.08%)    2,938 (100.00%)
  Fair                   |    950 (56.89%)     720 (43.11%)    1,670 (100.00%)
  Poor                   |    389 (53.36%)     340 (46.64%)      729 (100.00%)
Prior heart attack       |                                                    
  No heart attack        |  6,272 (63.53%)   3,601 (36.47%)    9,873 (100.00%)
  Had heart attack       |    275 (57.77%)     201 (42.23%)      476 (100.00%)
Diabetes status          |                                                    
  Not diabetic           |  6,233 (63.28%)   3,617 (36.72%)    9,850 (100.00%)
  Diabetic               |    314 (62.93%)     185 (37.07%)      499 (100.00%)
High lead level          |                                                    
  lead<25                |  2,902 (62.34%)   1,753 (37.66%)    4,655 (100.00%)
  lead>=25               |    209 (71.33%)      84 (28.67%)      293 (100.00%)
------------------------------------------------------------------------------

[Attaullah Shah]

For the mentioned task, both dtable and asdocx are excellent choices. While dtable allows for the flexibility of programming your table to your specific needs, asdocx offers simplicity and effectiveness (having fee of $9.99). The table you mentioned can be easily reproduced using the template(table1) feature in asdocx.

asdocx tab rural i.smsa i.sex i.race i.hlthstat i.heartatk i.diabetes i.highlead, ///
     template(table1) row replace save(table.tex)

Export to Excel?
Just change the file extension to .xlsx

asdocx tab rural i.smsa i.sex i.race i.hlthstat i.heartatk i.diabetes i.highlead, ///
     template(table1) row replace save(table.xlsx)

Similarly, you can export to Word (.docx) or HTML (.html) format. To know more about asdocx, you may like to visit https://asdocx.com

[Baker Chowdhury]

Jeff Pitblado (StataCorp) when using the survey data often time we need to use 'svy" could you please share the edits need for that ? Also, is it possible to get row proportion and standard error instead of confidence interval?

[Jeff Pitblado (StataCorp)]

Please see my post in this thread.