STATA labelling wrong?

[Roxanne Gray]

Hi all,

I have a ordinal variable with 4 levels, when I upload my data, STATA recognizes it as a string variable so I decided to code it as numeric and then label it properly. The first time I used this code it worked fine, but today for some reason STATA is labelling my variables wrong - and when I mean wrong, the frequency of each level is mixed up. I am not sure what to do!

This is my code:

encode category1, generate(ncategory1)
label define ncategory1 2 "Health maximiser" 1 "Pro-non-cancer" 4 "Strong prioritarian" 3 "Weighted prioritarian", replace

When I used the tabulate function before I obtained these results (Check image right.jpg)

but now I am obtaining results which are mislabeled and missing (check image wrong.jpg) PLEASE HELP

[daniel klein]

encode will always apply labels in alphabetic order. You must define the label first, then pass it to encode:

label define ncategory1 2 "Health maximiser" 1 "Pro-non-cancer" 4 "Strong prioritarian" 3 "Weighted prioritarian"
encode category1, generate(ncategory1) label(ncategory1)

This is documented in the help. Moreover, when you

label define ncategory1 2 "Health maximiser" 1 "Pro-non-cancer" 4 "Strong prioritarian" 3 "Weighted prioritarian", replace

you are changing the labels, not the underlying values. You never want that!


As for the different number of observations, you would need to provide a reproducible example. I cannot see how that could really happen given the same dataset.

[Roxanne Gray]

THANK YOU this worked!!!