PROC REPORT show overall estimates for both by groups

nd — Thu, 29 Nov 2018 15:27:17 GMT

I generated estimates of insurance status that I'd like to show by sex, race, and sex by race. Using PROC PRINT I can see the sex and race alone estimates, then sex by race. However, I want to split VARIABLE so that Uninsured and Insured are across. I can do this using PROC REPORT (see example below). However, I want sex alone to be grouped together as race alone is grouped before showing sex*race. Is there a way to do this using PROC REPORT?

PROC PRINT output

_ONE_ NEWRACE SEX VARIABLE pct low high flag

1	-2	-2	Uninsured	0.1956	0.1804	0.2117
1	-2	-2	Insured	0.8044	0.7883	0.8196
*	-2	Male	Uninsured	0.2328	0.2087	0.2587
*	-2	Female	Uninsured	0.1618	0.1432	0.1823
*	-2	Male	Insured	0.7672	0.7413	0.7913
*	-2	Female	Insured	0.8382	0.8177	0.8568
*	White, non-Hispanic	-2	Uninsured	0.1118	0.0926	0.1343
*	Black, non-Hispanic	-2	Uninsured	0.2389	0.2057	0.2755
*	Hispanic	-2	Uninsured	0.2944	0.2627	0.3282
*	Asian/PI, non-Hispanic	-2	Uninsured	0.1250	0.0896	0.1716
*	Other,non-Hispanic	-2	Uninsured	0.1550	0.0788	0.2824	*
*	White, non-Hispanic	-2	Insured	0.8882	0.8657	0.9074
*	Black, non-Hispanic	-2	Insured	0.7611	0.7245	0.7943
*	Hispanic	-2	Insured	0.7056	0.6718	0.7373
*	Asian/PI, non-Hispanic	-2	Insured	0.8750	0.8284	0.9104
*	Other,non-Hispanic	-2	Insured	0.8450	0.7176	0.9212	*
*	White, non-Hispanic	Male	Uninsured	0.1367	0.1076	0.1722
*	Black, non-Hispanic	Male	Uninsured	0.2927	0.2378	0.3544
*	Hispanic	Male	Uninsured	0.3656	0.3133	0.4214
*	Asian/PI, non-Hispanic	Male	Uninsured	0.1436	0.0928	0.2155
*	Other,non-Hispanic	Male	Uninsured	0.2298	0.0928	0.4653	*
*	White, non-Hispanic	Female	Uninsured	0.0861	0.0635	0.1156
*	Black, non-Hispanic	Female	Uninsured	0.1961	0.1579	0.2410
*	Hispanic	Female	Uninsured	0.2379	0.2004	0.2799
*	Asian/PI, non-Hispanic	Female	Uninsured	0.1045	0.0634	0.1676
*	Other,non-Hispanic	Female	Uninsured	0.1398	0.0697	0.2606	*
*	White, non-Hispanic	Male	Insured	0.8633	0.8278	0.8924
*	Black, non-Hispanic	Male	Insured	0.7073	0.6456	0.7622
*	Hispanic	Male	Insured	0.6344	0.5786	0.6867
*	Asian/PI, non-Hispanic	Male	Insured	0.8564	0.7845	0.9072
*	Other,non-Hispanic	Male	Insured	0.7702	0.5347	0.9072	*
*	White, non-Hispanic	Female	Insured	0.9139	0.8844	0.9365
*	Black, non-Hispanic	Female	Insured	0.8039	0.7590	0.8421
*	Hispanic	Female	Insured	0.7621	0.7201	0.7996
*	Asian/PI, non-Hispanic	Female	Insured	0.8955	0.8324	0.9366
*	Other,non-Hispanic	Female	Insured	0.8602	0.7394	0.9303

PROC REPORT output

Variable Uninsured Insuredsex race pct low high * pct low high *

-2	-2	0.1956	0.1804	0.2117		0.8044	0.7883	0.8196
	White, non-Hispanic	0.1118	0.0926	0.1343		0.8882	0.8657	0.9074
	Black, non-Hispanic	0.2389	0.2057	0.2755		0.7611	0.7245	0.7943
	Hispanic	0.2944	0.2627	0.3282		0.7056	0.6718	0.7373
	Asian/PI, non-Hispanic	0.1250	0.0896	0.1716		0.8750	0.8284	0.9104
	Other,non-Hispanic	0.1550	0.0788	0.2824	*	0.8450	0.7176	0.9212	*
Male	-2	0.2328	0.2087	0.2587		0.7672	0.7413	0.7913
	White, non-Hispanic	0.1367	0.1076	0.1722		0.8633	0.8278	0.8924
	Black, non-Hispanic	0.2927	0.2378	0.3544		0.7073	0.6456	0.7622
	Hispanic	0.3656	0.3133	0.4214		0.6344	0.5786	0.6867
	Asian/PI, non-Hispanic	0.1436	0.0928	0.2155		0.8564	0.7845	0.9072
	Other,non-Hispanic	0.2298	0.0928	0.4653	*	0.7702	0.5347	0.9072	*
Female	-2	0.1618	0.1432	0.1823		0.8382	0.8177	0.8568
	White, non-Hispanic	0.0861	0.0635	0.1156		0.9139	0.8844	0.9365
	Black, non-Hispanic	0.1961	0.1579	0.2410		0.8039	0.7590	0.8421
	Hispanic	0.2379	0.2004	0.2799		0.7621	0.7201	0.7996
	Asian/PI, non-Hispanic	0.1045	0.0634	0.1676		0.8955	0.8324	0.9366
	Other,non-Hispanic	0.1398	0.0697	0.2606	*	0.8602	0.7394	0.9303

proc report data=rse headline headskip  ;
	column (sex newrace) variable, (pct low high flag n) ;
	define sex / 'sex' group  order=internal ;
	define newrace / 'race' group order=internal;
	define variable / across nozero order=internal format=ins2f.;
	define pct / nozero format=7.4 ;
	define low / display format=7.4 ;
	define high / display format=7.4 ;
	define flag /'*'  ;
	define n / 'count' noprint;
run;

Thanks!

Re: PROC REPORT show overall estimates for both by groups

Cynthia_sas — Tue, 04 Dec 2018 22:07:47 GMT

Hi:
Nobody can run your code because you did not provide your format for "variable". Also, I'm having a hard time with your PROC REPORT output because I don't see how the code you've posted could have produced the output you've posted. The column headers don't look right. And, it's not clear to me why you have so many -2 values for NEWRACE and SEX. And this description of what you want "I want sex alone to be grouped together as race alone is grouped before showing sex*race." doesn't make sense to me. You have both SEX and NEWRACE in your column statement so that's what you'll get on the final report. If you want a report that shows sex only or a report that shows newrace only, then that would be a different PROC REPORT step because your column statement would have to change. Here's what I think you want (I used SASHELP.CLASSFIT to make some FAKE data.)

Here's the code I used:

data fakedata;
  length variable $10;
  set sashelp.classfit;
  where age in (12, 13, 14);
  do variable = 'Insured', 'Uninsured';
     output;
  end;
run;
  
proc report data=fakedata;
  column sex variable, (predict height weight);
  define  sex / group;
  define variable / across;
  define predict / mean;
  define height / mean;
  define weight / mean;
run;
  
proc report data=fakedata;
  column age variable, (predict height weight);
  define  age / group;
  define variable / across;
  define predict / mean;
  define height / mean;
  define weight / mean;
run;
 
proc report data=fakedata;
  column sex age variable, (predict height weight);
  define  sex / group;
  define age / group;
  define variable / across;
  define predict / mean;
  define height / mean;
  define weight / mean;
run;

Cynthia

topic Re: PROC REPORT show overall estimates for both by groups in SAS Procedures

PROC REPORT show overall estimates for both by groups

Re: PROC REPORT show overall estimates for both by groups