Mixed effects model

 The best way is to use PROC MIXED. Though you can use PROC GLM
 with proper care for random effects and fixed effects...

 First thing is of course to enter data is a proper way. Then
 comes the data analysis part.

 
 proc mixed data=syms method=type1 cl noinfo;
  class part oper;
  model meas=oper/cl outp=predout;  "predout" contains the predicted values
  random part part*oper ;
 run;

 proc print data=parmout;
 run;
 

 Here there are some important issues to discuss...
 1) note that the F-test based in any ANOVA should be in such
    a manner that under null H0 the statistics should follow
    a central F(a,b) distribution..

    Here we have the clear expressions of E(MSE) in all the
    cases which was very good for looking at this form..

 2) How to get the predicted response in this case.. "outp=predout"
    contains the predicted values and residuals.. but the point to
    be noted is that.. 

	residual = Y - YPred 

    and the predicted value is calculated using :

        YPred = X*b + C*V-1*(Y-X*b)

 Though these are not the desired residual but we can work with this to 
 how the model fits the data...