Document new and noteworthy features in version up to 2.8.3

[mhunter1]

I'll start

Handy S3 methods

• logLik to get the log likelihood from a model
• confint for Wald-type confidence intervals on free parameters
• anova for comparing nested models
• coef to extract free parameters
• simulate to generate data from a model

More functions

• mxFitFunctionMultigroup for multiple group models.
• mxFitFunctionMixture for (independent) mixture models
• mxFitFunctionHiddenMarkov for Markov dependent mixture models
• mxFitFunctionWLS & mxDataWLS for weighted least squares estimation
• mxGenerateData to generate data from a model
• mxSE to get standard errors on arbitrary algebraic expressions
• mxGetExpected to extract expected means, covariances, and/or thresholds from models
• mxMI to get modification indicies
• mxTryHard & mxTryHardOrdinal to repeatedly fit a model until it behaves
• mxAutoStart for automatic starting values
• omxDefaultComputePlan to generate a default compute sequence, for editing.
• mxBootstrap for nonparametric bootstrapping to get SEs and CIs
• mxParametricBootstrap for parametric bootstrapping to get marginal p-values for parameters
• mxComputeNelderMead for a general-purpose, derivative-free alternative to the GD optimizers
• imxRobustSE for robust standard errors in non-multilevel FIML analysis of raw data
• mxCompare for model comparison, which is now smart about detecting invalid comparisons, and which can do the bootstrap LRT
• mxCompareMatrix is a variation of mxCompare
• mxFactorScores to estimate regression, ML, or weighted ML factor scores
• mxCheckIdentification to check if a model is locally identified
• mxRefModels for fitting saturated and independence models

[RMKirkpatrick]

omxDefaultComputePlan() to generate a default compute sequence, for editing. mxBootstrap() for nonparametric bootstrapping to get SEs and CIs mxParametricBootstrap() for parametric bootstrapping to get marginal p-values for parameters mxComputeNelderMead() for a general-purpose, derivative-free alternative to the GD optimizers imxRobustSE() for robust standard errors in non-multilevel FIML analysis of raw data

[RMKirkpatrick]

mxCheckIdentification() to check if a model is locally identified. mxCompare() for model comparison, which is now smart about detecting invalid comparisons, and which can do the bootstrap LRT. mxCompareMatrix() is a variation of mxCompare().

[mhunter1]

I edited my original post to include @RMKirkpatrick 's suggestions.

[tbates]

perhaps close this one, and open a new item for next release?

[RMKirkpatrick]

perhaps close this one, and open a new item for next release?

OK with me.