Re: st: Polychoric PCA error message

[Yashin <yashin5@gmail.com>]

Dear Dr. Kolenikov,

Many thanks for your prompt and helpful response. I am still digesting
the details of your explanation below; what I understood more
generally is that while the Savalei paper recommends substituting 0.5
for zero in contingency tables as an adjustment in this procedure,
--polychoric-- does not include such an adjustment in its code. I hope
I understood correctly. If so, it appears that I have at least a
couple of options--to drop variables that have been leading to zeros
in contingency table cells, or to use another procedure, such as
ordinal PCA. Do you happen to know of other options I could consider?

Also, I encountered a situation that was unexpected for me. I
described in my previous email, and I copied it below--is this
surprising to you as well? If not might you recommend a reference that
might help me understand this better?

> 2) When I run the polychoric with only the dichotomous variables, and
> then with the same variables plus the additional 5 variables described
> above (ordinal and continuous), I get different correlation
> coefficients in the correlation matrix for the same variable pairs.
> How could this be? Sometimes the values are similar and yet different,
> and in other cases the values are quite different (some of the
> correlations > 0.9 when binary, ordinal and continuous variables are
> included in the matrix become zero when only binary variables are
> included in the matrix).

Thank you for your help,

Yashin

________________________________