topic Re: create summary percentage tables by subsets of one variables. in SAS Programming

create summary percentage tables by subsets of one variables.

Viveme789 — Fri, 12 Jan 2018 04:27:16 GMT

Updated

Hi! Suppose I have dataset as followings

ID SEX RACE AGE

1 F Non-hispanic White 5-15

2 M Asian/Other 5-15

3 M Asian/Other 15-20

4 F Hispanic 15-20

5 M Hispanic 15-30

6 M Non-hispanic Black 25-30

I want to create a table as following

	5-15	15-20	25-30
Sex
Male	50%	66.6%	100%
Female	50%	33.3%	0
Race
Non-Hispanic White	50%	0	0
Non-Hispanic Black	0	0	100%
Hispanic	0	66.6%	0
Asian/Other	50%	33.3%	0

I know how to use proc sql without sub-section age group but is there a way to aggregate all age sections in one step? Thanks!

Re: create summary percentage tables by subsets of one variables.

stat_sas — Fri, 12 Jan 2018 04:00:38 GMT

Hi,

Please try this.

proc freq data=have order=data;
table sex*age/norow nocum nofreq nopercent ;
run;

Re: create summary percentage tables by subsets of one variables.

Viveme789 — Fri, 12 Jan 2018 04:29:32 GMT

Thank you! I just updated my question. This problem is after running your code there is still a "total N" column and row. I need to generate all percentage tables using all variables with age, so my output need to be a dataset or table for future modification.

Re: create summary percentage tables by subsets of one variables.

Reeza — Fri, 12 Jan 2018 16:04:25 GMT

If this is for a report, consider using PROC TABULATE instead. You have more control over the output there, including percentages.

Re: create summary percentage tables by subsets of one variables.

ballardw — Fri, 12 Jan 2018 17:14:11 GMT

A report table could look something like:

proc tabulate data=have;
   class age sex race ;
   table sex race ,
         age=''*colpctn=''
         /misstext='0'
         ;
run;

add other categorical variables to the class and table.

Note that this will only work if all records have values for all of the variables as tabulate by default excludes records with any missing values for the variables on a class statement OR add /missing to the Class statement to get missing treated as a category.