generating a new var and replacing its values with the previous year's values

[Nader Mehri]

Hi everyone,
A sample of my data looks as follows (my current data ). As you can see, the person 1's starting health status (health_1) is H (healthy) in 2000 then becomes U (unhealthy) in 2002 and again H (healthy) in 2004. I would like to generate a new health variable (e.g., health_2) and replace it with the respondent health status in the previous wave.
Thanks.

Nader

*My current data

clear all
input id year str6 (health_1)
1 2000 "H"
1 2002 "U"
1 2004 "H"
2 2000 "H"
2 2002 "H"
3 2000 "H"
3 2002 "H"
3 2004 "H"
end
list

*My goal

clear all
input id year str6 (health_1 health_2)
1 2000 "H" ""
1 2002 "U" "H"
1 2004 "H" "U"
2 2000 "H" "H"
2 2002 "H" "H"
3 2000 "H" "H"
3 2002 "H" "H"
3 2004 "H" "H"
4 2005 ""  "new sample"
end
list

[Clyde Schechter]

Well, your proposed result is not consistent with your statement of what you want. How can health_2 be "H" for id 2 in 2000: there is no previous wave, so it should be missing. The same goes for id 3.

The following code gives what you said you wanted, not what you showed:

by id (year), sort: gen health_2 = health_1[_n-1]

Note: The above assumes that no id ever skips a wave.