data management/manipulation question

[Reese Crispen]

Hello,

In my dataset I have sets of variables that belong to different instruments. Currently, each observation contains data from each variable. However, I want each observation to contain only data from one instrument. Thus, for n instruments and m observations, I need nm observations in my new dataset. To illustrate:

I have:

obs	id	var1	var2	var3	var4
1	1	red	blue	dog	cat
2	2	yellow	green	mouse	bird

I want:

obs	id	instrument	var1	var2	var3	var4
1	1	colors	red	blue
2	1	animals			dog	cat
3	2	colors	yellow	green
4	2	animals			mouse	bird

Whats a nice way to go about this?

Thanks so much,
Reese

[Andrew Musau]

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte(obs id) str6 var1 str5(var2 var3) str4 var4
1 1 "red"    "blue"  "dog"   "cat" 
2 2 "yellow" "green" "mouse" "bird"
end

reshape long var, i(id) j(which)
gen which2= inlist(which, 1, 2)
egen id2= group(id which2)
reshape wide var, i(id2) j(which)

Res.:

. l

     +---------------------------------------------------------+
     | id2     var1    var2    var3   var4   id   obs   which2 |
     |---------------------------------------------------------|
  1. |   1                      dog    cat    1     1        0 |
  2. |   2      red    blue                   1     1        1 |
  3. |   3                    mouse   bird    2     2        0 |
  4. |   4   yellow   green                   2     2        1 |
     +---------------------------------------------------------+

[Reese Crispen]

Beautiful code Andrew. Thank you!