foreach results as a single table*

[Marcela Vieira]

Hi everyone,

I am calculating the percentiles of simulation data and would like to have the results as a single table instead of a table for each variable. I would like a single table so it is easier to make a graph.

- V2 to V21 are timesteps

Currently I am using this code:

foreach v of varlist v* {
centile `v' , centile(10 50 90)
}

The image below is a sample of what I am currently getting.

Any help is much appreciated,
Thank you

[Øyvind Snilsberg]

one way to do this,

tabstat v*, s(p10 p50 p90)

[Øyvind Snilsberg]

actually, if you want to make a graph maybe it is useful to save results to variables,

sysuse auto, clear
rename (price mpg rep78) (v1 v2 v3)

gen n = _n
gen c = .
gen lb = .
gen ub = . 

local k=0
foreach v of var v* {
centile `v', centile(10 50 90)
forvalues i = 1/3 {
replace c = r(c_`i') if n == `i'+`k'
replace lb = r(lb_`i') if n == `i'+`k'
replace ub = r(ub_`i') if n == `i'+`k'
}
local k=`k'+3
}
list c lb ub in 1/9, sep(3)

[Marcela Vieira]

I am still not having much success saving results to variables. I followed the command you recommended but when trying to list them I get the error "observation numbers out of range". The only thing I have changed is the below:

list c lb ub in 1/60, sep(3)

as I have 20 variables.

I believe the problem is because I only have 50 rows in my dataset (50 simulations) but not sure how to fix the problem. Could you please help me?

Many thanks again,

[Øyvind Snilsberg]

maybe add 10 empty observations at the end of the dataset,

insobs 10

gen n = _n
gen c = .
gen lb = .
gen ub = .

local k=0
foreach v of var v* {
centile `v', centile(10 50 90)
forvalues i = 1/3 {
replace c = r(c_`i') if n == `i'+`k'
replace lb = r(lb_`i') if n == `i'+`k'
replace ub = r(ub_`i') if n == `i'+`k'
}
local k=`k'+3
}
list c lb ub in 1/60, sep(3)