Foreach loop across variables

[Anushka Thachil]

Hi,
I have over a thousand dummy variables for 30,000 data entries. I want to find a way to count the number of times each one of those variables appears as 1 using a foreach 'variable' loop command. Would be helpful if there was another command. Here is the command I used:

After generating a new variable "fortier" to count the number of times a variable appears as '1', I then tried to replace it with a loop in the following way, where Stern and Pierneef are 2 out of the 1000 dummy variables.

foreach `var' of varlist Stern Pierneef {
replace fortier = count(`var') if `var'==1
}

[Niels Henrik Bruun]

Since we are in the Mata forum a quick way is:

. clear
. set rmsg on
r; t=0.00 8:17:53
. mata: // create dataset
------------------------------------------------- mata (type end to exit) ---------------
:     N = 30000
:     V = 1000
:     names = "v" :+ strofreal(1..V)
:     values = runiformint(N, V, 0, 1)
:     nhb_sae_addvars(names, values) // Install by ssc install matrixtools
: end
-----------------------------------------------------------------------------------------
r; t=64.84 8:18:58
. mata: //add a variable with rowtotals/rowcounts
------------------------------------------------- mata (type end to exit) ---------------
:     rowcounts = rowsum(st_data(.,.))
:     nhb_sae_addvars("rowcounts", rowcounts)
: end
-----------------------------------------------------------------------------------------
r; t=0.33 8:18:59
end of do-file
r; t=0.00 8:18:59

Note:

• I use my own Mata function nhb_sae_addvars() to insert variables and values into Stata
• The only thing You really need is line in red

[Nick Cox]

Meanwhile #1 is a little confused about local macros. Further, count() is an egen function, only.

But something like this should be closer to what you want. Naturally, your varlist will differ.

scalar sum = 0

foreach var of var Stern Pierneef {
    su `var', meanonly
    scalar sum = sum + r(sum)
}

where it is understood that an indicator variable is 0 or 1 or just possibly missing, so that the sum is identical to the number of 1s present.