Replacing variable values by the value of other variables

[Robyn Jones]

Hi all,

I have a panel dataset where i'm trying to create a variable that lists the month and year in which a given event first occured, however i'm struggling with the format of the date variables. I have variables for each individual wave of data collection that specify the month (iintdatmw*) and year (iintdatyw*) data was collected.

idauniq iintdatmw1 iintdatyw1 iintdatmw2 iintdatyw2
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005
100035 10 2002 2 2005

I have the data in long format so there are 9 observations for each wave of study, shown for id 100035 above. The data shows that in w1 this person was interviewed 10/2002 and in w2 interviewed 2/2005 – is there a way to create variables that would give the month and year also in long form? I have ~12000 obs in this dataset and have been wracking my brain over finding a solution, any help would be appreciated!

Many thanks,
Robyn

[Nick Cox]

Does this help?

gen ymw1 = ym(intdatyw1, intdatmw1) 
format ymw1 %tm

[Robyn Jones]

Hi Nick,

Thanks for the reply. This code has helped – i've done the following for each wave:

gen ymw1 = mdy(intdatmw1, day, intdatyw1)
format ymw1 %d

My data now looks like this:

idauniq ymw1 ymw2 ymw3 ymw4
100039 15may2002 15sep2004 15aug2006 15aug2008
100039 15may2002 15sep2004 15aug2006 15aug2008
100039 15may2002 15sep2004 15aug2006 15aug2008
100039 15may2002 15sep2004 15aug2006 15aug2008

whereas I need a variable that looks like this:

100039 15may2002
100039 15sep2004
100039 15aug2006
100039 15aug2008

Appreciate the help, thanks!

[Rich Goldstein]

h reshape

my guess as to the needed code (assumes these are the only variables):

reshape long ymw, i(idauniq) j(count)

you can replace "count" with whatever variable name you want