*
chapter_11_table_19

For the present (hypothetical) data set (taken from Shrout and Fleiss,
(1979) consists of six participants who are ranked by four judges. 

As is pointed out on page 563, the structure of Table 11.19 is 
analogous to that of Table 11.5 (repeated measures). However, notice 
that in the data file that the data is entered differently. For the 
repeated measures designs (e.g., 11.5), each row corresponded to a 
different participant while each column corresponded with another 
measurement. The main reason for the difference in how the data was 
entered is mainly because of the procedures used to analyze the data. 
SPSS and SAS allow repeated measures data to be entered in a 
"participants by occasions" fashion, whereas in order to get the 
appropriate mean squares to carryout the intraclass correlations for 
the data given in Table 11.19, a mixed effects (one fixed factor and
one random factor) ANOVA needed to be performed. It should be pointed 
out that the repeated measures data could have been entered in a similar
fashion as Table 19 and 29, however, the powerful repeated measures 
procedure in SPSS and SAS would not have been directly utilized. 

Because the intraclass correlations as given towards the end of Chapter
11 are not directly obtainable via SPSS or SAS, mixed effects ANOVAs 
are utilized such that the appropriate mean squares can be inserted 
into the appropriate formulas. Note that we obtain the mean squares by 
way of a mixed effects ANOVA, but either a random effects or a fixed 
effects ANOVA could have been utilized, because the corresponding mean 
squares are equivalent in each situation (the F ratios, which are not 
of interest here, are what is different).

The analysis proceeds by making use of Proc GLM, however because there 
is a random factor, a random statement is also included where the random
effect is placed to the right of the random statement.
;

LIBNAME md 'd:\data files by type\sas data files\tables';
     Data c11t19;
       SET md.chapter_11_table_19;
PROC GLM;
CLASS judge subject;
MODEL rating = judge subject judge*subject;
RANDOM subject;
RUN;