Counting observations with no missing data

[Van Ha]

Hi: I'm fairly new to STATA and was hoping to get some help. I have >100K observations with ~120 variables. Is there a code to count the number of observations with no missing data across the 120 variables? Thanks!

[Hemanshu Kumar]

egen miss = rowmiss(*)
count if miss == 0

[Van Ha]

That didn't work... count came back as 0. Is miss a new variable?

[Hemanshu Kumar]

Yes, miss is a new variable whose every observation contains the number of your existing variables that have a missing value in that observation. If the count is coming back as zero, that suggests that there are no observations that have no missing values on any variable -- there is something or the other that is missing in every observation. If you know this is untrue, perhaps you can start us off with a data extract using the dataex command, as suggested in the Statalist FAQ. That will help troubleshoot.

[Ken Chui]

Downloadable command mvpatterns can also tell you the missing pattern, and magnitude of missing for each variable:

net install dm91

sysuse auto, clear
mvpatterns make - gear_ratio

Results:

. mvpatterns make - gear_ratio
variables with no mv's: make price mpg headroom trunk weight length turn displacement gear_ratio

Variable     | type     obs   mv   variable label
-------------+---------------------------------------
rep78        | int       69    5   Repair record 1978
-----------------------------------------------------

Patterns of missing values

  +------------------------+
  | _pattern   _mv   _freq |
  |------------------------|
  |        +     0      69 |
  |        .     1       5 |
  +------------------------+

Just make sure the 120 variables are stored next to each other as a whole chunk, that way you can use a dash (-) to input them as an abstracted list, like:

mvpatterns variable001- variable120

Also, regarding thread #3, it's possible the count is 0, it just means that there is no case with no missing. If your 120 variables include opened ended questions as a follow up of "others", it's likely that you'd have 0. Try also tabulate miss and see the distribution of the other numbers.

[Nick Cox]

Ken Chui

There are several such and let's first note misstable as the quite versatile official command.

I note mdesc from SSC and missings from the Stata Journal, and I wouldn't be surprised at yet others.

[Ken Chui]

Ken Chui

There are several such and let's first note misstable as the quite versatile official command.

I note mdesc from SSC and missings from the Stata Journal, and I wouldn't be surprised at yet others.

Thanks, Nick. Noted. I'll take some time to get familiar with misstable.