Looping - Creating new variables

[Vinh Ng]

Dear all,

Suppose I have the following dataset.

input    var1    var2    var3    var4
12    2    3    12
4    5    6    24
23    65    9    54
21    7    12    512
2    8    16    12
6    10    10    12
end

I'd like to create new variables by dividing each variable (var1, var2, and var3) by a corresponding value of the variable var4.

More specifically, I'd like to create var5 by dividing var1 by the first observation of var4 (in this case, 12), var6 by dividing var2 by the second observation of var4 (24) and var7 by dividing var3 by the third observation of var4 (54).

Could anyone please help me with this ?

Thank you very much for your help,

Vinh

[William Lisowski]

Perhaps this code will start you in a useful direction.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(var1 var2 var3 var4)
12  2  3  12
 4  5  6  24
23 65  9  54
21  7 12 512
 2  8 16  12
 6 10 10  12
end
forvalues i=1/3 {
    local j = `i' + 4
    generate var`j' = var`i'/var4[`i']
    }
format var5-var7 %9.3f
list, clean noobs

. list, clean noobs

    var1   var2   var3   var4    var5    var6    var7  
      12      2      3     12   1.000   0.083   0.056  
       4      5      6     24   0.333   0.208   0.111  
      23     65      9     54   1.917   2.708   0.167  
      21      7     12    512   1.750   0.292   0.222  
       2      8     16     12   0.167   0.333   0.296  
       6     10     10     12   0.500   0.417   0.185

[Vinh Ng]

Hi William,

That is very helpful.

Many thanks,

Vinh

[Dung Le]

Dear William Lisowski,

You loop codes are really clever, however, could you please explain a bit about meanings of your codes? Say, at the 2nd code line, why do we need to put " + 4" and what is the meaning of adding
[`i'] right behind var4 at the 3rd code line? Thank you. DL

[William Lisowski]

These are the commands that the loop creates and runs.

    generate var5 = var1/var4[1]
    generate var6 = var2/var4[2]
    generate var7 = var3/var4[3]

We see that in the first command, i = 1 so j = 5 and var`i' becomes var1 and var`j' becomes var5 and var4[`i'] becomes var4[1] which is the value of var4 in observation 1.

[Dung Le]

William Lisowski, thank you so much for your detail explanations. I now understand how the loop commands work in this case.

Best regards,

DL