How to merge variables?

[Lange]

Hello!

I have three variables containing different things in the same dataset.

- Gender variable (v117): 0= male 1= female
- Employment variable (v246): 1= working 2=unemployed 3= pensioners etc etc.
- Job sector variable (v347): 0=private sector 1= public sector

These i need to be combined to a group which contains female from gender variable, the working from employment variable and the private from the job sector variable. This variable is those who are dependent on the welfare state versus the rest.

But how do i create a new variable where 0=depended on the welfare state and 1= in depended on the welfare state from these variables?

[Joseph Coveney]

I'm not sure what you mean by the welfare state or what you mean to be dependent upon it, but you can create variables using an if expression to qualify which observations get which value in the new variable being created, for example

generate byte mynewvar = v117 == 1 & inlist(v246, 2, 3) & v347 == 1

[Joseph Coveney]

I suggest starting with the list's FAQ, paying particular attention to Sections 9 and 12 (but don't overlook Section 6), and then take a look at the help files for Stata's logical operators, then generate and then if.

[Carlo Lazzaro]

Lange seems to want a new variable (let's call it - Deprivated people - ) that represents a union of different features specified in v117, v246 and v347.
I would figure out the following solution (that, however, I did not test):
-------------------------------------------------------------------------------
g Deprivated_people=1 if v117==1 | v246!=1 | v347==1
replace ~~Deprivated_people=0 if ~~Deprivated_people==.
------------------------------------------------------------------------------

Kind regards,
Carlo

[Dirk Enzmann]

I believe the best help you can get (by yourself) is

1. to follow the advice of Joseph (post #4)
2. to study the example given by Joseph (post #2) and Carlo (post #5) (thereby applying what you learned in step 1)
3. to generate the variable "independent" (or whatever you call it) they way you described in your last post (#6)
4. to test whether you really obtained the results you were looking for by using Stata's tab2 command with the option firstonly, for example:

tab2 independent gender employment jobsector, firstonly

You should also consider carefully whether the implicit definition of "independence" you proposed in your last post really makes sense. For example, should a woman employed fulltime as a public worker be classified as dependent (or depended)? But this is no longer a technical Stata question but a substantive decision of your own.

[Richard Williams]

how about something like this:

webuse lbw, clear
egen mygroup = group(low race smoke), l
tab1 mygroup

You can recode the variable however you want then.

[Nick Cox]

Cross-posted at http://www.talkstats.com/showthread....erge-variables

Please note our protocol on cross-posting in different forums in the Advice Guide (which also advises use of full real names).