Psychometric Mixture Models

Psychometric mixture models based on 'flexmix' infrastructure. At the moment Rasch mixture models with different parameterizations of the score distribution (saturated vs. mean/variance specification), Bradley-Terry mixture models, and MPT mixture models are implemented. These mixture models can be estimated with or without concomitant variables. See vignette('raschmix', package = 'psychomix') for details on the Rasch mixture models.


Changes in Version 1.1-7

o Set RNGversion("3.5.0") for reproducibility of previous results due to the changes in sample() for R version 3.6.0.

Changes in Version 1.1-6

o Load the "lattice" package in the vignettes in preparation to changes in the "flexmix" package.

Changes in Version 1.1-5

o Conditional registration of effect() and allEffects() methods when "effects" package is loaded.

Changes in Version 1.1-4

o Added a new function mptmix() for MPT mixture models, also known as latent-class MPT models. The design follows that of raschmix() and btmix(). Internally, the flexmix driver FLXMCmpt() is called to fit the finite mixture model. The function has not yet been fully tested and may change in future versions.

Changes in Version 1.1-3

o Updated the example for btmix.

Changes in Version 1.1-2

o The manuscript "Rasch Mixture Models for DIF Detection: A Comparison of Old and New Score Specifications." has now been published in Educational and Psychological Measurement, 75(2), 208-234. doi:10.1177/0013164414536183. A preprint version is included as vignette("scores", package = "psychomix").

o Improved functionality for raschmix() to allow for item response data in the itemresp-class.

o Improved axis labeling in plot function for raschmix objects.

o In the plot method nchar(..., type = "width") is now used to determine the default abbreviation.

o If suggested packages are needed internally, these are only called with :: semantics and not require()d anymore.

Changes in Version 1.1-1

o Adapted raschmix() to work with both the old psychotools version 0.2-0 and the new 0.3-0.

o Updated the "scores" vignette which is now also accepted for publication in Educational and Psychological Measurement.

Changes in Version 1.1-0

o Improved functionality for raschmix() to allow for differences between components in terms of identified parameters.

o Improved function raschmix() to leverage new functionality from the flexmix package: Parameter estimates from the previous M-step can now be used for initialization.

o Function raschmix() can now model the score distribution to be equal across all components (for both a "saturated" and a "meanvar" specification of the score model).

Changes in Version 1.0-0

o Official first stable release of the raschmix() functionality in the package, accompanying the manuscript "Flexible Rasch Mixture Models with Package psychomix" by Frick, Strobl, Leisch, and Zeileis, published in the Journal of Statistical Software 48(7). See citation("psychomix") for details.

Changes in Version 0.2-1

o For increased numerical stability the default minprior control parameter in raschmix() is now 0.05 (as in flexmix) and not 0 (as in the previous psychomix version).

o Revised vignette("raschmix", package = "psychomix"). Specifically, there is a discussion of how the FLXMCrasch() can be used directly with flexmix() or stepFlexmix() from the "flexmix" package.

o Improved function simRaschmix() to allow for a flexible specification of the data generating process.

o Added an effectsplot() function that leverages the "effects" package for visualizing the effects of the concomitant variables (if any) in the mixture model. This has not yet been fully tested and may change in future versions.

o Added a new function btmix() for Bradley-Terry mixture models. The design follows that of raschmix() rather closely. Based on from package "psychotools", there is a flexmix driver called FLXMCbtreg(). The btmix() function is a convenience interface calling stepFlexmix() with the FLXMCbtreg() driver. This has not yet been fully tested and may change in future versions.

Changes in Version 0.1-1

o First CRAN release of new "psychomix" package for fitting psychometric mixture models based on flexmix infrastructure. At the moment only Rasch mixture models are implemented in various flavors: with/without concomitant variables, different parametrizations of the score distribution (saturated vs. mean/variance specification). See vignette("raschmix", package = "psychomix") for details.

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


1.1-8 by Hannah Frick, 9 months ago

Browse source code at

Authors: Hannah Frick [aut, cre] , Friedrich Leisch [aut] , Carolin Strobl [aut] , Florian Wickelmaier [aut] , Achim Zeileis [aut]

Documentation:   PDF Manual  

Task views: Cluster Analysis & Finite Mixture Models, Psychometric Models and Methods

GPL-2 | GPL-3 license

Imports graphics, methods, stats, lattice, Formula, modeltools

Depends on flexmix, psychotools

Suggests effects, lmtest, nnet, numDeriv

See at CRAN