topic Re: SAS labeling help in New SAS User

SAS labeling help

lawye010 — Mon, 18 Jan 2021 23:58:42 GMT

Hello,

This is probably a very basic question but I was handed a large dataset set that does not have the variables numerically coded. I have imported the file but would like to relabel the variables with numerical codes. For instance for sex, right now the categories are male and female but I would like male=0 and female=1. How would I code this? Thanks.

Re: SAS labeling help

Reeza — Tue, 19 Jan 2021 00:06:17 GMT

It depends. How many variables do you have to do this for? How many different options....and most importantly, why?

Some applications require that you do this prior to your analysis for example, but SAS does not and can handle variables that are categorical in an analysis properly.

Some methods to fix this include:

Using custom informats and re-read the dat from the text file as needed
Using formats to remap the values to codes
Apply formats, changes how the variables appear but not the underlying data itself. Usually this is done in reverse, values are stored as numerics but displayed as text with full descriptions. Helpful for ensuring no typos and that the order is as desired
IF/THEN or SELECT statements to recode the values.

If you need to do the same remapping for multiple variables you may also want to use arrays.

Here's a tutorial on using Arrays in SAS
https://stats.idre.ucla.edu/sas/seminars/sas-arrays/

@lawye010 wrote:

Hello,

This is probably a very basic question but I was handed a large dataset set that does not have the variables numerically coded. I have imported the file but would like to relabel the variables with numerical codes. For instance for sex, right now the categories are male and female but I would like male=0 and female=1. How would I code this? Thanks.

Re: SAS labeling help

tarheel13 — Tue, 19 Jan 2021 01:03:42 GMT

proc format;

value sexf

0="Male"

1="Female"

;

run;

data dsn;

set dsn;

format sex sexf.;

run;

Re: SAS labeling help

ballardw — Tue, 19 Jan 2021 20:41:05 GMT

It may shortcut some question and answer cycles by running Proc Contents on your data set in question and sharing the result.

Another bit that may help is what do expect to do with 0/1 coding for sex that the current will not support?