Does not produce z-statistics for EFA models

[Ben Goodrich]

example(Factanal)

There are z-statistics for the SEFA model but not for the rotated EFA model.

[Ben Goodrich (bgokgm)]

There will be z-statistics if the raw data are passed to Factanal(), which allows a BHHH estimator of the asymptotic covariance of the estimates to be calculated. It seems to be a conservative estimate in small samples. But the method of Jennrich (1974) is planned for inclusion, which would not require the raw data necessarily.

[Ben Goodrich (bgokgm)]

There is now a completely overhauled mechanism for estimating the uncertainty of the estimates via simulation for all factor analysis models. Although this mechanism can be slow when calculating the uncertainty of rotated loadings in EFA models, it is valid in the sense that the confidence intervals take into account any inequality restrictions on (functions of) the parameters.