Generating dummy variables based on variables with multiple categories

[Matthijs Tichem]

Hi,

I have a (panel) dataset that contains the following variables for each observation: person_id - year - treatment.
Treatment is one of 65 categories, year: between 2012-2017 and person_id is a person identifier - unique to each person. Within a year person_id might appear more than once.

I would like to create a dummy variable which equals one for an observation if person_id appears more than once within the year of the observation, given that among the different observations there are at least 2 different treatment types.

If I do the following commands:

bysort person_id year: egen volume = count(treatment)
replace volume = 0 if volume == 1
replace volume = 1 if volume > 0

I get a dummy 'volume' which equals 1 for observations if the person_id appears more than once in the year of the observations. However, with these commands alone I cannot distinguish observations where person_id appears more than once in the year of observation and all treatment types are similar from those where there are at least 2 different treatment types.

Any ideas on how to do this?

Help is as usual much appreciated!

Thanks in advance.

[Clyde Schechter]

As you do not present example data, the following code is untested and may contain errors, but it should give you a start:

by personid year treatment, sort: gen treatment_count = (_n == 1)
by personid year: replace treatment_count = sum(treatment)
by personid year: replace treatment_count = treatment_count[_N]

This new variable, treatment_count will give you the number of distinct treatments that each personid has undergone in each year. You can then combine that with your volume variable to get what you want.

[Matthijs Tichem]

Dear Clyde,

Thanks for your reply! With the additional commands you provided, I was able to construct the variable exactly the way I wanted!