Greetings, Charles :-
Thank you for your comments on Marianne Belis' recent posting.
Marianne has just embarked on a trip to China, and will be gone
for a fortnight. I cannot speak for her anymore than she spoke for me
in her posting. Nevertheless, perhaps I can usefully comment in her
absence on one of the points you raise.
With respect to the provenance of weighted strength models, Marianne and
I have described hers (in the law-related work she mentioned in her posting)
as growing out of computational methods for geometric projections.
Wright's path analysis, which is also a kind of projection method,
relies on the chain rule, as you point out. While Marianne's proposal
admits of probabilistic interpretation, it does not implement the
probabilistic chain rule, nor does it otherwise represent a joint
probability distribution.
So, Wright and Marianne share a common ancestor and a common interest in
causality. She is not in his lineal descent. They differ from one another both
in form and in what aspects of causal relationships each describes.
Best regards.
Paul
