Prev: Saving the SAS log
Next: STATISTICAL ANALYSIS OF TEXT EMAILS Re: Refreshments after SOMS
From: "adel F." on 5 Dec 2009 11:07
=0AHi All,=0A =0AI would like to ask your advise on the following problem.=
=0A=0AI have the following=0Adesign=0A =0AThe filed was divided in two spl=
its (1 and 2)=0AThe are 50 blocks 43=0Ain split1 and 7 in split2.=0AIn ea=
ch block, the are two plots=0A =0AThe is an hierarchy design: Plots nested =
within blocks nested=0Awithin splits=0A =0AThere are 100 treatments. In eac=
h plot the treatment has been=0Aapplied randomly either with 1 replicate or=
2 replicates=0A =0AThe outcome is y (yield)=0A =0A =0APS: some plots have =
missing data=0A =0AThis is in my opinion an unbalanced incomplete design =
=0A =0AI suggested an anova with treatment and replicate (as fixed=0Aeffect=
variables), and block (as random effect variable), I ignored the split var=
iable for simplicity, I have used proc mixed below,=0A =0Aods output lsmean=
s =3Dlsmeans;=0Aproc mixed data=3Dmydata covtest noclprint;=0Aclass treatm=
ent plot=0Ablock replicate ;=0Amodel y=3D treatment replicate /solution;=
=0Arandom intercept /sub=3Dblock ; =0ALSMEANS treatment / CL ;=0Arun;=
=0A=0AI=0Awould appreciate your expert comments=0A=0AThanks a lot,=0A Adel=
=0A=0A=0A
 | 
Pages: 1
Prev: Saving the SAS log
Next: STATISTICAL ANALYSIS OF TEXT EMAILS Re: Refreshments after SOMS