list distinct values

[paulvonhippel]

I'd like to look at the joint distribution of 3 or more discrete variables a,b,c,...
I imagine a command called "list a b c, distinct" that would give output like this

a b c count
0 1 1 93
1 1 1 95
etc

where "count" is the number of rows with the given values of a,b,c.

In Unix, I can do this with "uniq -c". In SAS I can do it with PROC SQL, or with the LIST option in PROC FREQ tables.

How can I do it in Stata?

[Rich Goldstein]

either collapse or contract should do what you want

[paulvonhippel]

Thank you. Those commands change the data. Is there any way to produce the desired output without changing the data?

[Rich Goldstein]

yes, but it is rather more complicated:

qui count if a==0 & b==1 & c==1
local cat1=r(N)
qui count if a==1 & b==1 & c==1
local cat2=r(N)
etc.

you could, to make it easier, put it in a loop

then just list remembering to list the local also

[William Lisowski]

If you don't mind sorting the data, something like this might do what you need.

by a b c, sort: generate count = _N
egen tolist = tag(a b c), missing
list a b c count if tolist
drop count tolist

Caveat: I haven't tested this, and I'm uncertain how well it will work if any of the variables are missing.

[Nick Cox]

groups has been available to do this since 2003. The only real write-up is, I regret, now incomplete.

SJ-3-4 pr0011 . . . . . . . . Speaking Stata: Problems with tables, Part II
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N. J. Cox
Q4/03 SJ 3(4):420--439 (no commands)
reviews three user-written commands (tabcount, makematrix,
and groups) as different approaches to tabulation problems

However, the paper is dispensible. Just start with the help file if interested.

Here's an arbitrary example. (It is tidier in Stata than appears to be the case here.)

ssc desc groups
ssc inst groups

webuse nlswork

. groups race msp nev_mar collgrad not_smsa , sepby(race)

+--------------------------------------------------------------+
| race msp nev_mar collgrad not_smsa Freq. Percent |
|--------------------------------------------------------------|
| 1 0 0 0 0 1705 5.98 |
| 1 0 0 0 1 756 2.65 |
| 1 0 0 1 0 381 1.34 |
| 1 0 0 1 1 63 0.22 |
| 1 0 1 0 0 2300 8.07 |
| 1 0 1 0 1 717 2.51 |
| 1 0 1 1 0 739 2.59 |
| 1 0 1 1 1 189 0.66 |
| 1 1 0 0 0 7086 24.85 |
| 1 1 0 0 1 3761 13.19 |
| 1 1 0 1 0 1857 6.51 |
| 1 1 0 1 1 614 2.15 |
|--------------------------------------------------------------|
| 2 0 0 0 0 1268 4.45 |
| 2 0 0 0 1 391 1.37 |
| 2 0 0 1 0 157 0.55 |
| 2 0 0 1 1 29 0.10 |
| 2 0 1 0 0 1817 6.37 |
| 2 0 1 0 1 462 1.62 |
| 2 0 1 1 0 199 0.70 |
| 2 0 1 1 1 36 0.13 |
| 2 1 0 0 0 2304 8.08 |
| 2 1 0 0 1 932 3.27 |
| 2 1 0 1 0 402 1.41 |
| 2 1 0 1 1 42 0.15 |
|--------------------------------------------------------------|
| 3 0 0 0 0 17 0.06 |
| 3 0 0 0 1 4 0.01 |
| 3 0 0 1 0 2 0.01 |
| 3 0 0 1 1 1 0.00 |
| 3 0 1 0 0 50 0.18 |
| 3 0 1 0 1 6 0.02 |
| 3 0 1 1 0 28 0.10 |
| 3 0 1 1 1 4 0.01 |
| 3 1 0 0 0 108 0.38 |
| 3 1 0 0 1 36 0.13 |
| 3 1 0 1 0 37 0.13 |
| 3 1 0 1 1 10 0.04 |
+--------------------------------------------------------------+

[paulvonhippel]

groups is very nice, thank you!