Alternative display formats for same statistics in -dtable-

[Clyde Schechter]

I'm trying to create a table of descriptive statistics for a large data set. I have two variables for which I would like to present the 25th, 50th, and 75th percentiles. One of these is a date variable. (I know that it is unusual to present dates with these statistics, but I have my reasons in this case.) The other one is a count variable. So I would like the results for the date variable to be in %tdnn/dd/YY format, but the count variable in %1.0f. I have tried:

dtable, continuous(order_date, statistic(p25 p50 p75)) nformat(%tdnn/dd/YY) continuous(n_orders, statistic(p25 p50 p75)) nformat(%1.0f)

but that gives me everything as %1.0, so the dates are basically unreadable to humans. I also tried, pretty much knowing it wouldn't work:

dtable, continuous(order_date, statistic(p25 p50 p75) nformat(%tdnn/dd/YY)) continuous(n_orders, statistic(p25 p50 p75) nformat(%1.0f))

But, as I expected, I got

option nformat() not allowed
in option continuous()
r(198)

How can I do this?

* Example generated by -dataex-. For more info, type help dataex
clear
input float record_date int number_of_events
23100  2
23310  3
23140  4
22779  1
22741  7
23014  7
23198  4
23103  1
22882  1
22685  2
23267  1
22680  3
22987  3
22973  3
22707  1
23323  1
23124  2
23120  1
23071  3
23274 12
22718  1
22729  1
22660  1
23119 10
23022  2
23135  1
22683 12
23308  5
23219  2
23062  1
22942  1
23314  1
23035  2
22965  1
23125  1
22746  1
22986  1
23228  1
23251 15
23199  1
23099  5
22747  5
22629 19
22642  1
23187  1
23288  2
23168  1
23019  3
22647  1
23015  4
23114  4
23015  1
22673  7
22851 10
23287  1
23154  1
22715  5
23316  1
23316  2
22607  1
22690  2
22826  3
22767 11
23040  1
23126  1
23327  6
23323 59
23155  1
22612  1
22896  7
22698  3
22641  1
23115  1
23068  2
23215  3
23070  1
23224  1
23201  1
23285  1
22927  1
22984  8
23044  1
22951  1
22708  4
23091  8
23126  7
22939 23
22936  1
23070  2
23317 18
22783  2
22796  2
22847  9
22732  9
22752  2
22931  1
23178  3
23316  1
23236  1
22833  1
end
format %td record_date

[Andrew Musau]

The percentiles can also be denoted as q1, q2 and q3 - so you may use these in place of p25, p50 and p75. I am sure it is possible by defining some composite statistics, but my knowledge of dtable is still basic.

dtable, continuous(record_date, statistic(p25 p50 p75)) nformat(%tdnn/dd/YY p25 p50 p75) ///
continuous(number_of_events, statistic(q1 q2 q3)) nformat(%1.0f q1 q2 q3)

Res.:

. dtable, continuous(record_date, statistic(p25 p50 p75)) nformat(%tdnn/dd/YY p25
>  p50 p75) ///
> continuous(number_of_events, statistic(q1 q2 q3)) nformat(%1.0f q1 q2 q3)

----------------------------------------
                         Summary        
----------------------------------------
N                                    100
record_date      5/16/22 1/27/23 6/21/23
number_of_events                   1 2 4
----------------------------------------

[Clyde Schechter]

Thank you so much!!!