topic Re: Help combining levels of a categorical variable in Statistical Procedures

Help combining levels of a categorical variable

bazingarollcall — Thu, 18 Feb 2021 18:46:35 GMT

Hello everyone,

I've hit a wall with my own knowledge and am having a hard time finding the key words to search for topics to solve this problem via google or communities like this.

I have a categorical variable, plan_type, that is comprised of 9 levels, A, B, C, D, E, F, G, H, I. I want to combine B and C into a new variable to produce a proc freq table of the statistics. I also want to combine D, E, F, and G into one variable, and H and I into their own variable.

The goal is to produce frequency tables like this using proc freq and tables statements:

				Plan Type
		A	B	C	D	E	F	G	H	I	Total
Race/Ethnicity	White	10	5	2	456	156	486	123	4869	123
	Black	156	8469	5468	546	546	86947	31	84	320
	Hispanic	465	486	568	837	195	86	158	963	71
	Other	17	752	6382	5781	5714	72	85	185	81
Total

(the values here are not the values I'm working with and are just an example)

I have tried this code:

data health_data;
set health_data;
if plan_type=('A' 'B')
then variable='A_and_B';
run;

And it returned the error:

ERROR 180-322: Statement is not valid or it is used out of proper order.

Any ideas?

Thank you.

Re: Help combining levels of a categorical variable

PaigeMiller — Thu, 18 Feb 2021 19:14:20 GMT

You create a custom format, apply it to variable PLAN_TYPE, and then re-run PROC FREQ.

proc format;
    value $combf 'B','C'='BC' 'D','E','F','G'='DEFG' 'H','I'='HI';
run;

proc freq;
     /* whatever your code was before should go here, plus the following line */
    format plan_type $combf.;
run;

Re: Help combining levels of a categorical variable

bazingarollcall — Thu, 18 Feb 2021 19:32:37 GMT

Hi Paige,

This worked beautifully! Thank you, your solution also helped me learn how to use the $ operator.

I appreciate it!