topic Re: uniform gender values (M, Male, male, etc) in SAS Programming

uniform gender values (M, Male, male, etc)

fengyuwuzu — Wed, 02 Mar 2016 17:30:49 GMT

in my data, the gender has values like Male, M, m, male, Female, FEMALE, F, f.

When i use proc freq, they are all listed as different values.

I can recode them using if condition, but is there a simpler way to recode them unifomly?

Thanks

Re: uniform gender values (M, Male, male, etc)

FreelanceReinh — Wed, 02 Mar 2016 17:40:24 GMT

Hi @fengyuwuzu,

You could use character functions instead. Example:

data want;
set have;
length sex $1;
sex=lowcase(char(gender,1));
drop gender;
rename sex=gender;
run;

Re: uniform gender values (M, Male, male, etc)

Astounding — Wed, 02 Mar 2016 17:50:06 GMT

Another possibility: don't recode. Create a format to translate:

proc format;

value $gender 'M', 'm', 'Male', 'male' = 'Male'

'F', 'f', 'Female', 'female' = 'Female';

run;

proc freq data=have;

tables gender;

format gender $gender.;

run;

This leaves open the possibility of not changing your data, but still grouping the values for reporting purposes. It's a decent approach for cleaning data, as the PROC FREQ output displays values that occur in the data but are not accounted for using the format.

If you did want to change the values, you could still apply the format:

data want;

set have;

sex = put(gender, $gender.);

run;

Re: uniform gender values (M, Male, male, etc)

fengyuwuzu — Wed, 02 Mar 2016 18:04:56 GMT

Thank you, This also works!

Re: uniform gender values (M, Male, male, etc)

FreelanceReinh — Wed, 02 Mar 2016 18:35:11 GMT

@fengyuwuzu: When using the formatting approach you may want to use the order=formatted option in PROC FREQ statements, so as to ensure a consistent ordering of the two categories in the output. (Otherwise, the sort order would depend on the data because, for example, 'F'<'M', but 'M'<'f'.)

Re: uniform gender values (M, Male, male, etc)

ballardw — Wed, 02 Mar 2016 23:16:19 GMT

If I'm building the data set I often take @Astounding's idea one step earlie with an INFORMAT to read all of the values into my desired coding so later on all of the values a uniform.

Re: uniform gender values (M, Male, male, etc)

fengyuwuzu — Wed, 02 Mar 2016 23:18:19 GMT

Thank you for the tip, FreelanceReinhard.