How to recode categorial variables from 1,2,3 to 0,1,2?

[Ilse van Vliet]

Hi there,

I have a categorial variable that equals 1 if 'No', 2 if 'Yes, sometimes' and 3 if 'Yes, often'. Value 0 is not in use.

Now I want to recode this so that it equals 0 if 'No', 1 if 'Yes, sometimes' and 2 if 'Yes, often'.

Is there a certain command that obtains this goal?

Thanks,
Ilse

[Carlo Lazzaro]

Ilse:
see -help recode-.

[Maarten Buis]

The recoding is simple:

replace var = var -1

You will need to adept the the value labels, but that is just standard value label stuff. If you need help with that, please tell us.

[Ilse van Vliet]

Thank you Carlo and Maarten, this is helpful!

Maarten, I indeed might need some help with the value labels. This is what my data looked like before replace var = var-1:

halluc_freq |      Freq.     Percent        Cum.
------------+-----------------------------------
      never |     72,785       99.79       99.79
  sometimes |        137        0.19       99.98
  regularly |         16        0.02      100.00
------------+-----------------------------------
      Total |     72,938      100.00

And this is after:

halluc_freq |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |     72,785       99.79       99.79
      never |        137        0.19       99.98
  sometimes |         16        0.02      100.00
------------+-----------------------------------
      Total |     72,938      100.00

[Ilse van Vliet]

supplement to my previous post:

label define halluc_freq 0 "never" 1 "sometimes" 2 "regularly"

did not work.

I would be grateful for other tips!

[daniel klein]

See elabel (SSC, or https://github.com/kleindaniel81/elabel) for a canned solution

elabel adjust : replace var = var - 1

Done. Value labels are adjusted automatically.

[Nick Cox]

#5 You need the modify option too.

[daniel klein]

If you want to stay away from community-contributed commands, you first need to know the name of the value label attached to the variable halluc_freq. One way to find out is

describe halluc_freq

Before modifying or replacing the value label, I suggest making sure that no other variables are using the same value label. Perhaps something like:

ds , has(vallabel lblname)

But that still would not catch value labels in a different label language. elabel would run all these checks for you and tell you if there is something wrong. Anyway, sticking with core Stata, it might be better to create a new label from scratch:

label define halluc_freq_0_to_2 0 "never" 1 "sometimes" 2 "regularly"

then attach it to the (recoded) variable

label values halluc_freq halluc_freq_0_to_2

[Ilse van Vliet]

Daniel: Thank you for the elabel command, this was very helpful for variables coded as mentioned above. Your explanation in #8 was especially helpful for a categorial variable that equaled values 0; 101; 102; 201 and 202 (and now 0; 1; 2; 3; 4).

[daniel klein]

Your explanation in #8 was especially helpful for a categorial variable that equaled values 0; 101; 102; 201 and 202 (and now 0; 1; 2; 3; 4).

Here is one way to this with elabel:

elabel adjust : recode varlist (101 = 1) (102 = 2) (201 = 3) (202 = 4)