Tim Bates

``` mxCompareMatrix(models= c(fit1,fit2), statistic=c("minus2LL",'p')) One Factor Two Factor One Factor -3660.597 NA Two Factor NA -2456.24 > mxCompare(fit1) base comparison ep minus2LL df AIC diffLL diffdf p 1 One Factor...

unhelpful error: ```R mxCompareMatrix(base=fit1, models=fit2) Error in mxCompareMatrix(base = fit1, models = fit2) : mxCompareMatrix does not accept values for the '...' argument 'base' ``` Perhaps if given two models...

Just to note that this stops imxRobustSE working with twin models of the cont(top, MZ, DZ) pattern. https://openmx.ssri.psu.edu/comment/reply/4618/8794

Just to document current behavior: ```R m1 = umxRAM("WLS", data = mtcars, type="WLS", umxPath(c("wt"), to = "mpg"), umxPath("wt", with = "disp"), umxPath(var = c("mpg", "wt", "disp")) ) refModels = mxRefModels(m1,...

Just to note this is still broken so TLI etc. needing independence and saturated models can't be computed... As computing independence is just a single evaluation of `r'Wr = (s-sigma)'W(s-sigma)`...

Any chance to get `mxRefModels` working for WLS? The problem remains as it was in 2019: The user following the OpenMx instructions about getting fit indices is just told: >...

Hi mike. Hunter (above) described the implementation. Other apps implement and document the validity. But no, I can't implement this, but more than happy to help. For context, Hunter implementation...

fixed now: `.rda`

Looks like this is not doable - without perhaps making as as.boot=TRUE option to pass on some parts of the whole Model required for a boot object

We have RMSEA (and CI RMSEA) for model fit already... If the logic works (not sure it does) having an RMSEA target like