Describing missing data

[Matthew Breckons]

Hi, I'm a new Stata user and I've gone around in circles a bit with this. I'm basically trying to find a percentage of missing data over multiple variables. I'm only after a specific missing value (.a) which I've used to label unexpected missing. I can use 'tab var1, missing' to see the percentage of '.a' values in a single variable but how do I go about viewing the percentage of '.a' values within the whole dataset? I.e. so that I can report the overall level of missing data.

Any help much appreciated.

Matt

[Clyde Schechter]

Try this:

local dot_a_count = 0
ds, has(type numeric)
local numeric_vars `r(varlist)'
foreach v of local numeric_vars {
    count if `v' == .a
    local dot_a_count = `dot_a_count' + r(N)
}
display "Proportion of .a values among numeric variables: _continue"
display =`dot_a_count'/(`=_N' * `:word count `numeric_vars'')

Warning: not tested, beware of typos

[Matthew Breckons]

Thank you for this.

Do you know if there is a way of getting a single overall figure? A total percentage of missing values within these variables?

[Joseph Coveney]

What Clyde recommends gives you that, doesn't it? At least when you multiply it by 100 to convert proportion to percentage.

Edited: If you're looking for any missing value, then change

count if `v' == .a

to

quietly count if missing(`v')

[Matthew Breckons]

Apologies - you're right - it does, I had misinterpreted the output. Thanks so much, it really is appreciated!

Could I ask one further question - if I'm trying to narrow this down to obtain the figure for a group of specified variables - which part(s) of the code doI replace with these? So if I'm after the proportion of missing data withing var 1, var 4 & var6?

[Oded Mcdossi]

just list your variable under the ds command

[Matthew Breckons]

Thanks for your help all.