Creating an indicator variable if one event occured within plus/minus 15 days within another event.

[Jan Kovar]

Dear Statalisters,

I have perhaps a rather basic question I do not know the answer to. I have a list of events with a particular date (in my case European Council summits) captured in one variable and a list of events with another date (in my case occurrences of politicians' tweets). I would like to create a thirnd varibales( dichotomous), which would capture whether a tweet has occurred with plus/minus 15 days of an European Council summit.

This is an example from my dataset:

* Example generated by -dataex-. For more info, type help dataex
clear
input float(European_Council Tweet_date Tweet_Summit)
17776 17794 .
17820 17829 .
17843 17939 .
17877 17585 .
17957 18067 .
17975 18074 .
17992 18085 .
18066 17947 .
18157 18137 .
18199 18167 .
18220 18174 .
18241 18209 .
    . 18249 .
    . 18249 .
    . 18308 .
end
format %td European_Council
format %td Tweet_date
label values Tweet_Summit Yes_No

Achieving it is basically easy using the following:

replace Tweet_Summit=1 if inrange(Tweet_date, dmy(17,8,2008), dmy(16,9,2008))
(0 real changes made)

However, with hundreds of occurrences of one event (European Council summits) this would be time-consuming as one would have to manually calculate plus/minus 15 days for each summit and input it into the command.

Is there a more elegant solutions?

Many thanks, and best wishes
Jan

[Clyde Schechter]

gen `c(obs_t)' obs_no = _n
gen low = European_Council - 15
gen high = European_Council + 15
rangestat (count) wanted = obs_no, interval(Tweet_date low high)
replace wanted = 0 if missing(wanted)
replace wanted = min(wanted, 1)
replace wanted = . if missing(European_Council)

-rangestat- is written by Robert Picard, Nick Cox, and Roberto Ferrer. It is available from SSC.