Struggling with sorting categorical data

JSMI1443 — Wed, 01 Mar 2023 03:44:15 GMT

I am trying to sort data that has 4 possible categorical data options: Independent, Republican, Democrat, (blank).

I am trying to make a confidence interval, but first I need to have the categorical variables be numeric characters instead, so I tried to sort them using:

data=gss08_Q1;
input=gss08;
if polparty = 'independent' then polparty1 = 1;
if polparty = 'rebublican' then polparty1 = 2;
if polparty = 'democrat' then polparty1= 3;
else polparty= 4;

This returned with multiple of the same errors, am I using this wrong?

Re: Struggling with sorting categorical data

Patrick — Wed, 01 Mar 2023 05:07:16 GMT

You've got other syntax errors in your code like data=gss08_Q1; but I'm not going to comment on a code snippet with no error log provided.

Your "core" syntax should work like tested below once you fix the last assignment

Adding a few more ELSE for efficiency wouldn't hurt as well.

data test;
  polparty='independent';
  if polparty = 'independent' then polparty1 = 1;
  ELSE if polparty = 'rebublican' then polparty1 = 2;
  ELSE if polparty = 'democrat' then polparty1= 3;
  else polparty1= 4;
run;

Another approach for recoding a value is using a format/informat. Especially if you've got a lot different values for a category this leads to code that's easier to read.

In below example using an informat with the upcase option you've got also the advantage that using the format will work on source strings whether upper-, mixed- or lowcase.

proc format;
  invalue pp_num(upcase)
  'INDEPENDENT' =1
  'REPUBLICAN'  =2
  'DEMOCRAT'    =3
  other         =4
  ;
run;

data test2;
  polparty='independent';
  polparty1=input(polparty,pp_num.);
run;

topic Struggling with sorting categorical data in SAS Procedures

Struggling with sorting categorical data

Re: Struggling with sorting categorical data