Proc Report: Across scattered

[Tpham]

I am new to proc report. Using across to nest two variables together. However, all the results are not on the same row and are scattered

This is what I have:

THis is my code and sample data structure 

Data have;
retain IGA group	Week1	Week2	CatC	CatN;;
length CatC $ 10.;
input IGA group	Week1	Week2	CatC $	CatN;
cards;
1	3	.	.	N	1
1	3	.	.	Mean	2
2 	3	1	1	N	1
2 	3	2.3	2.0	Mean	2
3	3	3	3	N	1
3	3	1.9	2.0	Mean	2
4	3	3	3	N	1
4	3	0.9	1.3	Mean	2
5	3	4	4	N	1
5	3	1.7	1.3	Mean	2
;

ods tagsets.rtf file="&path.\output\Draft\&pgm..rtf" style=rtfdef startpage=yes nogtitle nogfootnote;

proc report data=have nowindows headline headskip SPANROWS split='|' spacing=4 
	
		style(header)={ just=center asis=on} 
		style(column)={asis=on} 
		style(report)={width=9 in};

	column group CatN CatC (IGA, (Week1 Week2));

	define IGA/across order=data style(header)=[just=center fontweight=bold  asis=on] style(column)=[just=center asis=on];
		
	define group/order=data group "Age Group" style(header)=[just=left fontweight=bold  asis=on] style(column)=[just=left asis=on cellwidth=0.7in];

	define CatN/order=data noprint;
	define CatC /display 'Statistic' style(header)=[just=center fontweight=bold  asis=on] style(column)=[just=left asis=on cellwidth=0.7in];

	define Week1/display  style(header)=[just=center fontweight=bold  asis=on] style(column)=[just=center asis=on cellwidth=0.7in ];
	define Week2/display  style(header)=[just=center fontweight=bold  asis=on] style(column)=[just=center asis=on cellwidth=0.7in ];
	
run;

ods tagsets.rtf close;

How do I get all N and mean to report on the same row across the IGA variable and weeks?

[Tpham]

I found this article https://www.lexjansen.com/pharmasug/2012/TF/PharmaSUG-2012-TF20-SAS.pdf 

I've tried the dummy variable method and that did not work.

[Tpham · Accepted solution]

Closing this as I figured it out. The had to change week variable from display to analysis

[ghosh]

Your data structure is not the same as in the article, where they use a single variable for gender (not separate M and F). 

To use the code in the paper,  I would suggest you reorganize your data in a longitudinal timeseries format, with a single variable representing week number. 

This will also give you the flexibility to add additional weeks when needed.