proc glm with mean differences

[starz4ever2007]

anyone know how to get the mean difference by week using proc glm:

so far I have the code:

proc glm data=work.psilongfinal ;
class week abc_scr ;
model new_total = week*abc_scr / ss1 ss2 ss3 ss4 ;
means week*abc_scr;
format abc_scr abc_scr.;
run;

and I get this: SAS Output

		N	new_TOTAL
WEEK	abc_scr	Mean	Std Dev
1	1	96	12.4913	6.41615
1	2	50	16.5664	7.28447
8	1	120	7.55877	5.49867
8	2	10	12.6476	5.18959
12	1	118	6.59475	5.24013
12	2	9	11.5119	6.38991
16	1	118	6.81316	6.03729
16	2	7	7.9	5.34516

I would also like:

		N	new_TOTAL
WEEK	abc_scr	Mean	Std Dev
1-8	1		12.49-7.52=
1-8	2		16.57-12.65=
1-12	1		12.49-6.54=
1-12	2		16.57-11.51=
1-16	1
1-16	2

 

[starz4ever2007]

sorry the values in the second table should be in the "Mean" column not the "Std Dev"

[PGStats · Accepted solution]

proc glimmix gives you more options. This little example might inspire you:

proc glimmix data=sashelp.cars plots=none;
where cylinders in (4,6) and make in ("Chrysler", "Ford", "Chevrolet") ;
class make cylinders;
model MSRP = make|cylinders;
lsmeans make*cylinders / slicediff=cylinders diff;
run;

[starz4ever2007]

was able to modify and make it work! thanks 😄