equatiomatic::extract_eq() method for "nestedLogit" objectsextract_eq() when dichotomy names contain "_"submodel= arg to avoid $submodel indexingRSQ() function to compute pseudo-R² measures (McFadden, Cox-Snell, Nagelkerke, and others) for each dichotomy sub-model and for the combined polytomous modelRSQ() in the nestedLogit and other-examples vignettesRSQ.multinom() and RSQ.polr()This is a major release of the package adding considerable functionality for plotting and other features
gators data, food choice of alligators from Agrestiggeffects package for nestedLogit modelseasystats packages for nestedLogit models [not yet a vignette, because work on this is continuing]predict.nestedLogit() and related to its own file for improved documentationplot.nested-ci.Rplot.nestedLogit() to use scales::hue_pal() for consistency with ggplotscale argument to plot.nestedLogit() to allow plotting on the logit (log-odds) scale (scale = "logit") in addition to the default probability scale (scale = "prob")as.tree() method for ASCII printing of the tree of nested dichotomiesGSS data, "l.t.highschool" changed to "<highschool"lty not passed to matplot() in plot.nestedLogit()plot.nestedLogit(), added label.col argument (defaulting to the value of col in the call), to control the color of the curve labelggeffects vignetteggeffects vignette with predict_response() for nestedLogit models; see ggeffects issue #671other-examples vignette traced to tibble not handled well in plot.nestedLogitrhub::rhub_check() workflow.This is a minor release, improving documentation
pkgdown documentation in DESCRIPTIONcreateDichotomies methods now explicitly internalEffect method for "nestedLogit" objects to create effect displays.as.data.frame method when newdata has one columnpredict.nestedLogit() now includes the newdata data frame as an additional component (.data) in its result. Consequently, the newdata argument is no longer required in the as.data.frame method.This is a major enhancement to the package, adding computations of standard errors and confidence intervals to predicted probabilities and logits.
predict() method now computes standard errors for probabilities and logits using the delta method. These can be obtained for either the response probabilities (and equivalent logits) or for the predicted log odds of the individual dichotomies.confint() method for the result of predict() generates the corresponding confidence intervals.as.data.frame() methods for predicted values, either for the nested logit model or for the separate dichotomies converts these to a data frame in long format, handy for using ggplot().confint() method for predicted probabilities and logitsplot() method for "nestedLogit" objects now plots confidence intervals for predicted probabilities.predict() methods and focusing exclusively on `ggplot2().lobstr::tree() to print nested listsfig.show="hold" to keep par(op) with the code.par() and options() calls so as to not alter user's workspace.plot() method now operationallinearHypothesis() method to give Wald tests for hypotheses about coefficients or their linear combinations.models() generic and method to extract separate models from the "nestedLogit" objectlogLike() method, and through it, gets AIC() and BIC()NEWS.md file to track changes to the package.