Destring Command (STATA)

Sometimes, you will find strings variable from STATA dataset. Unfortunately, you cannot do summarize, regression, ANOVA, graph, etc. However, sometimes, the dataset is saved as strings. This is a character variable, then you should to change this variable to numeric value, in order to be calculated by the program.

Simple question, how to change strings variable to numeric variable?

. d b1r1 b1r2 b1r3 b1r4 b1r5 b1r7 b1r8

storage display value
variable name type format label variable label
——————————————————————
b1r1 str2 %9s
b1r2 str2 %9s
b1r3 str3 %9s
b1r4 str3 %9s
b1r5 str1 %9s
b1r7 str7 %9s
b1r8 str2 %9s

The variables above are the strings variable. You cannot do calculation, eg:

. sum b1r1 b1r2 b1r3 b1r4 b1r5 b1r7 b1r8

Variable | Obs Mean Std. Dev. Min Max
————-+——————————————————–
b1r1 | 0
b1r2 | 0
b1r3 | 0
b1r4 | 0
b1r5 | 0
————-+——————————————————–
b1r7 | 0
b1r8 | 0

You can change to numeric variable. You can type this command:

destring b1r1 b1r2 b1r3 b1r4 b1r5 b1r7 b1r8, generate(b1r1_new b1r2_new b1r3_new b1r4_new b1r5_new b1r7_new b1r8_new) ignore(“$ ,%”)

. destring b1r1 b1r2 b1r3 b1r4 b1r5 b1r7 b1r8, generate(b1r1_new b1r2_new b1r3_new b1r4_new b1r5_new b1r7_new b1r8_new) ignore(“$ ,%”)
b1r1 has all characters numeric; b1r1_new generated as byte
b1r2 has all characters numeric; b1r2_new generated as byte
b1r3 has all characters numeric; b1r3_new generated as int
b1r4 has all characters numeric; b1r4_new generated as int
b1r5 has all characters numeric; b1r5_new generated as byte
b1r7 has all characters numeric; b1r7_new generated as long
b1r8 has all characters numeric; b1r8_new generated as byte

The result is here:

. d b1r1 b1r2 b1r3 b1r4 b1r5 b1r7 b1r8 b1r1_new b1r2_new b1r3_new b1r4_new b1r5_new b1r7_new b1r8_new

storage display value
variable name type format label variable label
—————————————————————————————————
b1r1 str2 %9s
b1r2 str2 %9s
b1r3 str3 %9s
b1r4 str3 %9s
b1r5 str1 %9s
b1r7 str7 %9s
b1r8 str2 %9s
b1r1_new byte %10.0g
b1r2_new byte %10.0g
b1r3_new int %10.0g
b1r4_new int %10.0g
b1r5_new byte %10.0g
b1r7_new long %10.0g
b1r8_new byte %10.0g

Yes, you’ve got numeric variable! Summarize, then you get:

. sum b1r1_new b1r2_new b1r3_new b1r4_new b1r5_new b1r7_new b1r8_new

Variable | Obs Mean Std. Dev. Min Max
————-+——————————————————–
b1r1_new | 277854 42.28353 24.82338 11 94
b1r2_new | 277854 22.01755 26.53184 1 79
b1r3_new | 277854 63.87709 62.68575 10 730
b1r4_new | 277854 8.370929 8.581939 1 132
b1r5_new | 277854 1.571617 .4948454 1 2
————-+——————————————————–
b1r7_new | 277854 2069906 319858.5 1500101 2599001
b1r8_new | 277854 5.461858 2.858474 1 10