Carrying forward values under conditions of another variable

[Jeffery Sauer]

Dear Statalist,

Suppose I have the following dataset:

input state dist year value
1 1 1990 10
1 1 1991 20
1 1 1992 30
1 1 1993 40
1 1 1994 50
1 1 1995 60
1 1 1996 70
1 1 1997 80
1 1 1998 90
1 1 1999 100
1 1 2000 110
end

My object is to use certain values for different sets of years (of unequal length). For example, I would like to use the 1990 value for all years between 1990-1993, the 1994 value for all years between 1994-1995, and the 1996 value for all years between 1997-2000.

While at first I thought I could use a replace with an [_n-#] modifier, the unequal time spans (and sometimes missing start/end points) are proving challenging. Is there an alternative to replace I am missing?

Best,

Jeff

[Clyde Schechter]

I assume you want to do this -by- state and dist.

gen new_value = value if inlist(year, 1990, 1994, 1996)
by state dist (year), sort: replace new_value = new_value[_n-1] if missing(new_value)

[Jeffery Sauer]

Clyde to the rescue again! To think of all those you have helped.

Not sure why my attempt with -bysort- was going awry. Thank you very kindly!

[Nick Cox]

https://www.stata.com/support/faqs/d...issing-values/ gives general discussion.