Generating a new variable but missing values have been included

[Joe Tuckles]

Hi,

My code is:

generate var_66 = 1 if var_33 >=2

However it has put a 1 in the new variable where there are missing values in var_33, rather than just where there is a score of 2 or greater in var_33.
Any advice?
Thanks

[Carlo Lazzaro]

Joe:
Stata considers -.- as the highest number (hence your condition does not qualify for missing values exclusion).
A possible fix would be:

replace var_66==.
generate var_66=1 if var_33 >=2 & var_33!=.

[Joe Tuckles]

Perfect thanks for the clarification

[Nick Cox]

In Stata {1, 0} or {1, 0, missing} indicators are often more useful than {1, missing} indicators. So typically I would go

generate var_66 = var_33 >= 2 if var_33 <  .

Notes: < . is shorter than but equivalent to != . if there are no extended missing values. If there are extended missing values, it excludes them too.

[William Lisowski]

In logical expressions like x>=2 Stata evaluates the expression to true when x is missing, treating missing values as greater than any nonmissing value, a consequence of the way Stata stores missing values. This differs from arithmetic expressions, where x+2 would be missing if x were missing. Furthermore, you code will generate var_66 as a missing value when var_33<2, which s not a good idea. Either it should be 0/1; or it should be 0/1/missing.

So what you need is something like the following

generate var_66 = var_33 >=2 & var_33<.

which will generate a 0 if var33 is <2 or missing, and a 1 otherwise.

Or perhaps you would prefer to have the missing value propagate to the new value, in which case the following will help

generate var_66 = var_33 >=2 & var_33<.
replace var_66 = var_33 if missing(var_33)