Re: suggestions for software?

Tony Downing (
Mon, 20 Oct 1997 14:37:28 +0100

Dear Janie,

Re your message:
>I am doing a piece of research for my MSC dissertation on the uses of Rep
>Grid technique and am looking back through abstracts over several years. I
>am principally interested in whether the grid has been used for research
>purposes or for the benefit of clients.
>I have a grid of data consisting of 137 elements and 5 constructs rated on
>a scale of 0-3. I have analysed and clustered it by hand but am wondering
>if there is a program which could cope with that number of elements which I
>could use to validate my analysis?

Oneobvious thing you could do would be to carry out a principal components
analysis (PCA) on the correlations between the constructs. It may be that
any of the usual grid packages would be fine for this. If not, you could
use any standard stats package, e.g., Minitab, SPSS of SAS. The much
greater number of elements than constructs would be a positively good
thing, from the point of view of PCA. Of course with only 5 constructs you
may not be wanting to claim to have sampled adequately the range of
constructs that might bave been in the possible repeertoire, but, if they
cover a narrow domain, perhaps you have, and although there are only 5 of
them it may still be of interst to know, for example, how much to the
variabnce overall is accountable for in terms of, say, 1, or 2 or 3
construct-factors. And if 1 or 2 factors did account for most of the
construct variance, you could meaningfully plot factor scores for all those
137 elements in factor space generated by that number of factors, as a way
of visualising the similarities and differences among those elements.

Tony Downing, M.A., Ph.D.

Lecturer in Psychology, Dept.of Psychology,University of Newcastle upon Tyne,
Ridley Building, Claremont Place, Newcastle upon Tyne, NE1 7RU, England.
Telephone: +44 (0)191 222 6184, Mobile: +44 (0)468427481
Fax: +44 (0)191 222 5622