CHAPTER 11

                 COMPUTING CORRELATION MATRICES 



  The  option  for  computing a correlation matrix enables you to 
compute a correlation matrix containing up to  51  variables.   A
raw  data  file  must  be used as the input data file.  The input 
data file may have up to 200 variables and there is no  limit  to
the number of cases.  If you use a raw data file having more than
51  variables,  you must select 51 or fewer of them for inclusion
in the correlation matrix.


                      REORDERING VARIABLES

  The  correlation  matrix program will present an opportunity to 
re-order your variables if you wish to do that.  If you elect  to
re-order  your  variables,  complete  instructions for doing that
will be presented to you at that time.


                         MISSING VALUES

  The program enables you to select either a listwise or pairwise
option for processing missing values.


                        LISTWISE DELETION

  The listwise deletion option is nearly  always  preferred  when
computing  correlations.   If  that option is chosen, the program 
will retain and use only those  cases  for  which  there  are  no
missing  values  for the variables that are to be included in the
correlation matrix.


                        PAIRWISE DELETION

  The  use  of  the  pairwise  deletion  option  when   computing
correlations  will always produce correct correlations.  However, 
if the  resulting  correlation  matrix  is  then  used  in  other
procedures such as multiple regression or factor analysis, it can
produce  misleading  or  incorrect  results.   Thus, the pairwise
deletion option should always be used with great caution and then 
only when you have a relatively small proportion of missing  data
within a respectably large sample.

  The  treatment  of  missing  values  is  discussed at length in 
Chapter 7 of the  text  by  Cohen,  J.   &  Cohen,  P.   'Applied
Multiple   Regression/Correlation  Analysis  for  the  Behavioral
Sciences', (2nd ed). Hillsdale, NJ: Lawrence Erlbaum, 1983.

  Because of the huge demands on memory, the pairwise deletion of 
missing values is not included in  the  current  version  of  the
multiple   regression   program.    However,   you  can  use  the 
correlation matrix program  to  produce  correlations  using  the
pairwise  deletion  option.   If  you  will  save  the  resulting 
correlation matrix on a disk file it can then be  read  into  the
multiple  regression  program  using  the  'summary  data'  input
option.


                       TESTING THE PROGRAM

  If you wish to test the program to see how it functions, choose
option  'B'  from  the main menu and then use the data file named 
LONGLEYR.DAT that is stored on your SPPC Disk 1.  Be sure to  use
a missing values code of -1 or 1e200.