Creating a variable that only contains initial year of implementation

[Garam Chu]

Hello all,

I have a quick question, and it might be straightforward. (However, I could not figure it out in an efficient way)
My data has STATEID, YEAR, TREATMENT, and other covariates. What I wanted to do was to create a relative year. So, first, I wanted to create a variable that includes only the first start year of the treatment.
For example, "GA" is treated from 2000 through 2020, but "FL" is treated from 2005 through 2018. Then, the new treatment variable for GA should be only equal to 1 if the year is 2000 and 2005 for FL (0 for all other years).
May I ask how I can solve this problem?

Thank you!

[Øyvind Snilsberg]

bysort stateid (year): gen wanted = year if treatment
bysort stateid (wanted): replace wanted = wanted[1]

[Garam Chu]

Hi,

Thank you for your assistance! But, may I ask if I could create a variable that is equal to 1 if that is the initial treatment year and all others equal to 0?

[Øyvind Snilsberg]

* Example generated by -dataex-. For more info, type help dataex
clear
input str2 stateid int year byte treatment
"GA" 1999 0
"GA" 2000 1
"GA" 2001 1
"GA" 2002 1
"FL" 2003 0
"FL" 2004 0
"FL" 2005 1
"FL" 2006 1
end

egen x = min(year) if treatment, by(stateid)
gen wanted = x == year

list stateid year treatment wanted, clean

       stateid   year   treatm~t   wanted  
  1.        GA   1999          0        0  
  2.        GA   2000          1        1  
  3.        GA   2001          1        0  
  4.        GA   2002          1        0  
  5.        FL   2003          0        0  
  6.        FL   2004          0        0  
  7.        FL   2005          1        1  
  8.        FL   2006          1        0

[Nick Cox]

@Øyvind Snilsberg is bang on here, but for a very recent thread with more discussion and links see

https://www.statalist.org/forums/for...st-observation

[Garam Chu]

Thank you all! It worked and I appreciate your help again.