CORRTOT computes the pooled means, standard deviations, and a pooled
correlation matrix of the combined sample given the sample sizes, means,
standard deviations and correlation matrices of two independent samples
(based on listwise deletion of missing cases). To compute the pooled
correlation matrix, the the sample sizes, means and standard deviations
of the two samples to be combined must be known. CORRTOT handles a
maximum of 49 variables.

CORRTOT interactively asks for the number of items and the names of the
two files, each containing the following data in the following sequence:

1) One or more rows of the means of the variables

2) Subsequently, one or more rows of the standard deviations of the
   variables

3) Subsequently, a symmetric (full) correlation matrix of the variables

4) A last row containing the sample size

The data of 1) to 3) may be given on several rows, however, each type
of data (means, standard deviation, correlation coefficients, sample size)
must start with a new row. Of course, each file must contain the data of
same number variables.

Example of two files containing data for 3 variables:

Input file 1:

 12.5 13.453 .98
  8.2 11.4 .6034
1.000 0.421 0.123
0.421 1.000 -0.346
0.123 -0.346 1.000
 48

Input file 2:

 9.2 14.541 1.02
 8.2 9.3 .5034
1.000 0.395 0.098
0.395 1.000 -0.288
0.098 -0.288 1.000
 72

Output file contains the pooled data of the two input files:

 10.5200000000 14.1058000000  1.0040000000
  8.3252940494 10.1596507199  0.5434916996
  1.0000000000  0.3859804965  0.0994049832
  0.3859804965  1.0000000000 -0.3140235464
  0.0994049832 -0.3140235464  1.0000000000
 120

The following SPSS-syntax can be used to produce such input files:

-------------------------------------------------------------------------------

CORR item01 to item20,item31
  /missing listwise
  /print nosig
  /matrix=out('\spsswin\jobs\corrs.mat')
  /statistics descriptive.

GET '\spsswin\jobs\corrs.mat'.
if $casenum=1 rowtype_='AMEAN'.
if $casenum=2 rowtype_='BSTDDEV'.
if $casenum=3 rowtype_='ZN'.
execute.
sort cases by rowtype_.
execute.
PRINT outfile='\spsswin\jobs\corrs.cor'
  /item01 to item20,item31.
EXECUTE.

-------------------------------------------------------------------------------

The following SPSS-syntax can be used to read the output of CorrTot, for
example to compute a factor analysis of the data of the combined sample:

-------------------------------------------------------------------------------

MATRIX DATA variables = item01 to item20,item31
     /FILE = 'c:\spsswin\jobs\corrtot.cor'
     /FORMAT FREE FULL
     /CONTENTS=mean sd COR n_scalar.

FACTOR
  /MATRIX = IN(cor=*)
  /ANALYSIS item01 to item20,item31
  /PRINT EXTRACTION ROTATION
  /CRITERIA FACTORS(3) ITERATE(50)
  /EXTRACTION=PAF
  /ROTATION VARIMAX .

-------------------------------------------------------------------------------

If you want to combine more than two samples, you can run CorrTot repeatedly,
using the output of the first run as an input file in the second etc.

To facilitate the combination of the data of more than two samples, there is
a DOS version of CorrTot (file = CorrTotD.exe). The DOS-version can handle
data sets of up to 50 variables. You can call the DOS-version with four
parameters:
 1) number of Variables (=Items);
 2) name of input-file containing the data of a first sample;
 3) name of input-file containing the data of a second sample;
 4) name out output-file containing the data of the combined sample.

For example:
CORRTOTD 3 SAMPLE1.COR SAMPLE2.COR TOTAL.COR

The DOS-version can be run by using a batch-job in which CorrTotD is called
repeatedly, see the example file CorrComb.BAT. Thus, the number of samples
to be combined is only limited by the space of your harddisk.