In the spirit of recent requests for things to be included in version 8, I'd
like to add better factor and correspondence analysis and MDS. At the
minimum, make factor and rotate external, with easy access, so users can add
new factoring procedures without, for example, having to write their own
varimax rotation. Among the features I would like are:
1. The ability to iterate communalities only a user specified number of
times, consistent with Gorsuch's claim that iterating 3-4 times is optimal.
2. Orthogonal and oblique Procustes rotation. This should be done in such
a way that one could use Procrustes rotation with other data than factor
results.
3. Better access to factor results, so one can easily manipulate them.
4. Built in correspondence analysis, updating coranal and mca, as these
don't seem to work well with version 7, often producing errors. Also, more
options.
5. Multidimensional scaling (MDS), including 3-way (INDSCAL) models.
6. I'd love to have 3-way principal components analysis, a la Kroonenberg,
but I realize this is too much to expect.
These are exploratory data procedures and/or data reduction procedures that
psychologists use and are not available to a level that we are used to. Of
these, 1, 3, & 4 should be easy and are the minimum. In other words,
improving some of these features to the level of the wonderful
model-building features.
Cheers,
Stephen Soldz
Boston Graduate School of Psychoanalysis
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
