Help on labeling categorical variables

[Geralt Ji]

Hello

I'm working with my data file but find difficulty on labelling derived categorical variables.
In my data set, there are more than 100 crude variables and they range from 1 to 28.
And crudecat variables are derived from crude variables.
Then, according to crude variables, I want to label my variables as following rules:

crude variables		crudecat variables
if crude variables=1	→	1
if crude variables=2	→	2
if crude variables=3	→	3
if crude variables>3	→	4+

It's easy to do that at first, I type:
gen crudecat=.
replace crudecat=1 if crude==1
replace crudecat=2 if crude==2
……

But when I entered "replace crrudecat=4+ if crude>3 & crude<3", it didn't work. And I also noticed that "gen" command only create numerical variables rather than categorical variables that I want.

So I tried to use "label" command, I typed:
label define crudecat 1 "1" 2 "2" 3 "3" >3 "4+"

or

label define crudecat 1 "1" 2 "2" 3 "3" 3/99 "4+" (Because the range from crude variables are 1 to 28 and I guessed "/" can use to denote "range to" , but it failed. After that command, my crude variables which equal to 4 change to "/" and others stay same. I don't know why.)

or
label define crudecat 1 "1" 2 "2" 3 "3" 3/max "4+"

All methods show errors and don't work. Most of time we only label variables with fixed number like:
label define crudecat 1 "1" 2 "2 "

It tortures me for days.
Thanks for your time and help

[daniel klein]

It seems that values 1 to 3 are really just copied. So the following should suffice to create the variable(s) that you want.

generate crudecat = crude
replace  crudecat = 4 if crude >= 4

Likewise, there is no point in assigning label "1" to 1 and so forth; simply

label define crudecat 4 "4+"
label values crudecat crudecat

should do.

Before you ask your next question, please take some time to review the FAQs. Data examples help; use CODE delimters (as I have above) to make code (and output) better readably; statements, such as "did not work" or "showed error" are imprecise and do not help diagnosing the problem, better include a carbon copy of what you typed and what Stata did (or did not) in response.

[Geralt Ji]

Thanks Daniel Klein！