Degree of freedom for Period effect in TYPE III

LAYMAN_YO — Fri, 22 Sep 2017 04:47:49 GMT

data ex01;
input period sequence subject treat carryover y;
cards;
1 1 1 1 0 50
1 1 2 1 0 55
1 2 1 2 0 44
1 2 2 2 0 51
1 3 1 3 0 35
1 3 2 3 0 41
1 4 1 1 0 54
1 4 2 1 0 58
1 5 1 2 0 50
1 5 2 2 0 55
1 6 1 3 0 41
1 6 2 3 0 46
2 1 1 2 1 61
2 1 2 2 1 63
2 2 1 3 2 42
2 2 2 3 2 45
2 3 1 1 3 55
2 3 2 1 3 56
2 4 1 3 1 48
2 4 2 3 1 51
2 5 1 1 2 57
2 5 2 1 2 59
2 6 1 2 3 56
2 6 2 2 3 58
3 1 1 3 2 53
3 1 2 3 2 57
3 2 1 1 3 57
3 2 2 1 3 59
3 3 1 2 1 47
3 3 2 2 1 50
3 4 1 2 3 51
3 4 2 2 3 54
3 5 1 3 1 51
3 5 2 3 1 55
3 6 1 1 2 58
3 6 2 1 2 61
;

proc glm data=WLEX01;
CLASS SUBJECT SEQUENCE TREAT PERIOD CARRYOVER;
MODEL Y=SEQUENCE SUBJECT(SEQUENCE) PERIOD TREAT CARRYOVER ;
LSMEANS TREAT /PDIFF;
RANDOM SUBJECT(SEQUENCE);
RUN;

Please explain why the TYPE III degree of freedom for PERIOD is "1" , not "2" ?

Re: Degree of freedom for Period effect in TYPE III

SAS_Rob — Mon, 25 Sep 2017 15:54:15 GMT

It looks like period=1 and carryover=0 are aliased

proc freq data=ex01;
tables period*carryover;
run;

Re: Degree of freedom for Period effect in TYPE III

LAYMAN_YO — Tue, 26 Sep 2017 01:23:54 GMT

Thank you very much for the answer.

topic Re: Degree of freedom for Period effect in TYPE III in Statistical Procedures

Degree of freedom for Period effect in TYPE III

Re: Degree of freedom for Period effect in TYPE III

Re: Degree of freedom for Period effect in TYPE III