Re: PCA analysis of rep grids

Richard Bell (
Thu, 13 Mar 1997 09:59:21 +1100


At 07:00 AM 3/11/97 +1000, you wrote:
>Regarding your reply:
>>G-Pack is to be replaced by Gridstat in the next few months.
>Would you like to give an indication of what is on it/how it differs from

It is meant to be a replacement for G-pack. It looks very similar on the
outside but is more compact and better organized on the inside. The
clustering is the same (de-bugged), the eckart-young (singular value)
decomposition factor analysis has been modernized (leaping from a 1965
fortran II algorithm of Horst to a 'modern' 1982 one which produces 'nicer'
scaled loadings), and correlation matrices are rescaled if necessary to
allow squared -multiple-correlations to be calculated as better indices of
'intensity'. Added is principal components of construct correlations with a
range of possible rotations. I'm open to suggestions about what else might
go in. At the moment I'm planning to talk about it in Seattle.


Richard C Bell
Department of Psychology
University of Melbourne
Parkville Vic 3052 Australia
Phone: +61 (0)3 9344 6364
Fax: +61 (0)3 9347 6618