dtable sorted

[Sarah Heavey]

Is there a way to generate a table using dtable, with rows sorted in descending order? Essentially what's achieved with

tab var , sort

.

[Kevin Blaine]

I've got the same question. Did you ever find a solution?

[Simon Pfaff]

Hallo,

If I understand you correctly, you could use "myaxis" to reorder a categorical variable. Here is a short example:

webuse auto, clear

* using tabulate
tab rep78, sort

* Reorder categorical variable
myaxis rep78_sorted = rep78, sort(count) descending

* using dtable
dtable i.rep78
dtable i.rep78_sorted

. dtable i.rep78

-----------------------------
                     Summary
-----------------------------
N                          74
Repair record 1978           
  1                  2 (2.9%)
  2                 8 (11.6%)
  3                30 (43.5%)
  4                18 (26.1%)
  5                11 (15.9%)
-----------------------------

. dtable i.rep78_sorted

-----------------------------
                     Summary
-----------------------------
N                          74
Repair record 1978           
  3                30 (43.5%)
  4                18 (26.1%)
  5                11 (15.9%)
  2                 8 (11.6%)
  1                  2 (2.9%)
-----------------------------

[Kevin Blaine]

Awesome -- thank you!

[Kevin Blaine]

Hi Simon -- back again! How might this work across multiple variables, if at all?

For example, I have 5 binary variables, and I want a single row for each variable's "Yes" frequencies/percentages. The following code for dtable works a charm for this:

unab temp_symptoms: sym*
        foreach symp of local temp_symptoms {
            local symptoms = "`symptoms' 1.`symp'"
            }

        dtable num_symp `symptoms', ///
            by(path, notests) nformat(%9.1fc p50 p25 p75) ///
            title(Table 1c. Symptomatology by Tumor Type) titlestyle(font(,size(10)))    ///
            notestyle(font(,size(6)))
        fix_dtable_total path
        collect style cell, font(,size(8))
        collect style header `temp_symptoms', level(hide)
        collect style putdocx, layout(autofitc) halign(center)

However, the sort order is based on the order the variables were placed in the local `symptoms'. How could I sort a local containing variables by the frequency of a particular value within each variable?

[Kevin Blaine]

Of course, I figured out the answer minutes after posting. For anyone else with a similar problem

vorter

from SSC did the trick. Here's my implementation below:

unab temp_symptoms: sym*    
foreach symp of local temp_symptoms {
            local symptoms = "`symptoms' 1.`symp'"
            }
vorter - (sum) `temp_symptoms', after(num_symp)

This reorders the variables in my dataset, but since they are grouped together anyway, the order doesn't matter to me.