Dummy variable doesn't work, what am I doing wrong?

[Jonathan Smits]

Hello,

I want to generate a dummy variable that gives a value of 1 if the value of a variable is between the range of the 5th and 95th percentile, 0 otherwise


I did the following:

gen X_inrange = inrange(X, r(p5), r(p95))

But afterwards for all firm-year observations it is giving an 1... I do not observe any 0s.. It gives a 1 for every observation no matter what...


What am I doing wrong?

[Carlo Lazzaro]

Jonthan:
you may want to try:

. set obs 1000

. g X=runiform()+1000

. centile X, centile(5 95)

                                                       -- Binom. Interp. --
    Variable |       Obs  Percentile    Centile        [95% Conf. Interval]
-------------+-------------------------------------------------------------
           X |     1,000          5    1000.037        1000.027    1000.046
             |                   95     1000.94        1000.931    1000.957

. return list

scalars:
             r(n_cent) =  2
                  r(N) =  1000
               r(ub_2) =  1000.957302458706
               r(lb_2) =  1000.930621496683
                r(c_2) =  1000.939984130859
               r(ub_1) =  1000.045955191923
               r(lb_1) =  1000.027411419193
                r(c_1) =  1000.036810302734

macros:
           r(centiles) : "5 95"

. gen X_inrange = inrange(X, r(c_1), r(c_2))

[Nick Cox]

It's only going to work while those results remain visible. Note that @Carlo applies generate immediately after the quantile calculation.


Otherwise Stata substitutes missings and that means your calculation will always return 1 for true:

. di inrange(-1e9, ., .)
1

. di inrange(1e9, ., .)
1

[Jonathan Smits]

Thanks guys!