Work on this project was started around 1985 by Henny Coolen and
Abby Israels. It is incomplete and fragmentary. At the same time
maximum likelihood based OS methods were implemented by Maarten
van Wijk and Frits Meijerink. This can be compared with the
linearization methods of De Leeuw and with the polychoric
methods of Muthen and Lee.
Extend to latent variable models,
and have it include factor analysis (FACTALS). The idea that
a latent variable is a variable with all observations missing.
