forvalues + pctile

[Leonie Schneider]

I am trying to compute 99th percentile value across distinct values of a given variable. Basically, i am trying to generate a variable "p99_depth" which shows the 99th percentile value for values of depth corresponding to each individual instrument. My knowledge of STATA is fairly limited and this is the code i came up with:

//i generated a new variable 'group' to number each distinct value across my variable of interest (instrument):

egen group = group(instrument)
su group, meanonly
forvalues i = 1/`r(max)' {
gen p99_depth = pctile(depth), p(99) if group == i
}

This is giving me invalid syntax error. Any help fixing the code or an alternative approach to solving this problem would be appreciated. Thanks

[Nick Cox]

Four errors that I can see:

1. Should be egen not gen.

2. The if qualifier is in the wrong place.

3. The local macro lacks quotation marks.

4. Second time round the loop, p99_depth already exists and the loop would fail.

But as pctile() can be coupled with by: you don't need a loop at all.

bysort instrument: egen p99_depth = pctile(depth), p(99)

[Leonie Schneider]

that worked, thanks a lot!!