*    This is file KIDNEY.SPS 
TITLE 'Two way ANOVA with missing data' 
DATA LIST FILE=kidnymis.dat LIST 
        / A Hyper N1 N2 Nephron 
VAR LABELS A='Na-K-ATPase activity' 
           Hyper='normal or hypertensive rats' 
           Nephron='nephron site' 
*  MANOVA requires consecutive integer values for group codes, so we 
*  need to recode the Hyper factor 
RECODE Hyper(-1=0) 
MANOVA A by Hyper (0,1) Nephron (1,3) 
      /Design=Hyper Nephron Hyper by Nephron 
*  The default sums of squares are SSTYPE(UNIQUE) which are the 
*  marginal sums of squares need for missing data (what SAS GLM calls 
*  the 'TYPE III' sums of squares). 
* 
*  Now, do a missing cell problem (no interaction) 
DATA LIST FILE=kidmiscl.dat LIST 
        / A Hyper N1 N2 Nephron 
VAR LABELS A='Na-K-ATPase activity' 
           Hyper='normal or hypertensive rats' 
           Nephron='nephron site' 
RECODE Hyper  (-1=0) 
MANOVA A by Hyper (0,1) Nephron (1,3) 
  /Error=W+R 
  /Design=Hyper Nephron 
*  The ERROR=W+R statement pools the residual and within cell errors to obtain 
*  the error term necessary to conduct tests based on the marginal sums of 
*  squares.  This command only needs to be entered if there are is one or more 
*  empty cells and interactions left out of the model.  If there are no empty 
*  cells and all interactions are in the model, including the ERROR=W+R * 
command will not affect the residual sum of squares. 
FINISH