Recursive partitioning based on psychometric models, employing the general MOB algorithm (from package partykit) to obtain Bradley-Terry trees, Rasch trees, rating scale and partial credit trees, and MPT trees.
Changes in Version 0.15-2
o The method behind mpttree() has now been published by Behavior Research Methods, see doi:10.3758/s13428-017-0937-z
o The predict() method for "bttree" objects works again correctly for type = "parameter" (previously called type = "worth") and type = "rank". Thanks to Heather Turner for pointing out the problem.
o node_btplot() used to fail for worth=FALSE or ref= but works now correctly again. Thanks to Heather Turner for pointing out the problem. Also ref= now either matches to $labels or (if not available) to colnames() to facilitate usage with the PlackettLuce package.
o node_regionplot() now also works correctly on devices with a default white (rather than transparent) background such as png or jpeg.
o node_* panel functions gained bg = "white" argument to enable different background filling color.
Changes in Version 0.15-1
o The recently added mpttree() function is now accompanied by a working paper, see citation("psychotree") or ?mpttree for details.
o Properly imported grDevices in NAMESPACE.
o Basil Abou El-Komboz changed his name to Basil Komboz.
Changes in Version 0.15-0
o New function mpttree() and visualization function for model-based recursive partitioning of multinomial processing tree (MPT) models. These functions are somewhat experimental, and their user interface might change in future releases.
Changes in Version 0.14-0
o Internals of package psychotree completely reorganized to employ the recent additions and changes in psychotools (>= 0.3-0) and to use the new partykit (>= 0.2-0) implementation of the mob() function (rather than the old party implementation).
o The handling of argument 'ref' when producing a region/effect plot was changed. Whereas in the previous implementation, the restriction specified in this argument was applied to the cumulative absolute item threshold parameters, it now is applied to the absolute item threshold parameters.
o New panel-generating visualization functions 'node_profileplot' and 'node_regionplot' have been added which replace the old functions 'node_raschplot' and 'node_effects'.
Changes in Version 0.13-0
o New functionality for recursive partitioning of partial credit and rating scale models via functions pctree() and rstree(), respectively. These require package "psychotools" >= 0.2-0 which provides the corresponding basic model fitting tools.
Changes in Version 0.12-3
o Improved Depends/Imports/Suggests declarations in DESCRIPTION.
Changes in Version 0.12-2
o Adjusted vignette 'raschtree' according to the slightly different results due to the new approximation formulae for p values in the supLM test used in party::mob() (>= 1.0.3).
o Improved the package's tests by adding reference output for the examples and vignette for comparison checks.
Changes in Version 0.12-1
o Moved all base functionality to the new "psychotools" package in order to provide common infrastructure for both "psychotree" and "psychomix" as well as other packages. This affects the "paircomp" class and associated methods as well as the model fitters btReg.fit() and RaschModel.fit() and associated methods.
o The data sets Firstnames, GermanParties2009, and Soundquality have also been moved to psychotools.
Changes in Version 0.11-1
o Enhanced raschtree() functionality, introduced in detail in the new vignette("raschtree", package = "psychotree").
o Added subsample from SPIEGEL Studentenpisa (SPISA) quiz for students from Bavaria that answered the same set of questions.
o Added artificial data set DIFSim exhibiting differential item functioning, employed for illustrating Rasch trees.
Changes in Version 0.9-1
o Added raschtree() function, based on simple RaschModel() (for binary 0/1 items).
o Some bug fixes (subsets with missing drop = FALSE) for paircomp() and bttree() with only two objects
Changes in Version 0.9-0
o First CRAN release of the package accompanying the forthcoming paper "Accounting for Individual Differences in Bradley-Terry Models by Means of Recursive Partitioning" by Strobl, Wickelmaier, and Zeileis, accepted for publication in Journal of Educational and Behavioral Statistics.
o Compared to earlier package versions available from R-Forge, this package does not depend on data/code from the "prefmod2" package. All required functionality (basic Bradley-Terry regression, paircomp class, data sets) are now contained in the "psychotree" package.