In "SAS System for Mixed Models" by R. Littell et al., on page 495 there is this code:

proc mixed;

class a b block;

model y = a|b;

random block a*block;

run;

This is for a split-plot design involving two experimental factor, A and B, and two different sizes of experimental units to which they are applied. The levels of A are randomly assigned to the larger size of experimental unit, called whole plot, whereas the levels of B are assigned to the smaller size of the experimental unit, the subplot. The subplot are assumed to be nested wihtin the whole plot so that the whle plot consists of a cluster of subplots and a level of A is applied to the entire cluster. The classical example is the application of fertilizer to large plots of land, and different crop varieties planted in subdivisions of the large plot. For this example fertilizer is the whole plot factor A and variety the sub^plot factor B.

Dear Jos Jos Feys

Thank you for your comment. I mean a factorial split-plot design within years not a simple split plot. I found a code to analyze but not sure about that.

Here the code:

data a;

input year r a b c y;

cards;

.

.

.

;

proc print;

run;

proc glm;

class year r a b c;

model y=year r year*r a year*a b year*b a*b year*a*b r(a b) c year*c a*c b*c year*a*c year*b*c a*b*c year*a*b*c/ss3;

test h=year e=year*r;

test h=a year*a b year*b a*b year*a*b e=r(a b);

means a b/lsd e=r(a b);

means c/lsd;

run;

proc sort;

by A;

proc glm;

class r B C;

model y=r B C;

means B C/LSD;

by A;

run;

proc sort;

by A B;

proc glm;

class r C;

model y=r C;

means C/LSD;

by A B;

run;

Hi Hossein,

I think that my first question would be: Did the plots and subplots were the same for both years? There are two alternatives, to consider year as a random effect or as a fixed effect. Considering year as a fixed effect, you just add the variable year in the model, as a split-split plot design (I have the code for it). Let me know if my answer helped or if you need further information.

Regards,

Hi

The answer is yes. They were same. Also, year should be random.

This is the code useing proc mixed if you consider as a random effect (if you don't have a block factor, just take it out from the model)

proc mixed method = type3;

class block plot subplot year;

model total_weight = block year plot|subplot / outp = new1 ddfm = satterth;

random block(year) block(year)*plot ;

lsmeans plot|subplot / pdiff;

run;

Pedro Rojas-Barros Thank you. I will run it.

check this from google

SAS Code for Some Advanced Experimental Designs