0.7.1

• The algorithm warns about and handles singular matrices in initial values of the inner iteration loop.
• The convergence criterium is based on the change in the linear part of the model instead of the deviance, like MASS::glmmPQL

0.7

• The algorithm has been completely refactored. It works much more along the lines of MASS::glmmPQL
• Internally a Cholesky-factor parametrisation is used to prevent negative variance estimates

0.6.3

• mblogit now handles responses with only two columns.
• mblogit now can deal with matrix responses that have rows that sum to zero.
• mclogit and mblogit now return a component named "df.residual" instead of "residual.df". (Fixes issue #4)

0.6.2

Merged PR from pmcharrison - typo fix fft -> fit

0.6.1

• Fixed matrix column selection in predict.mclogit if there is only one
  predictor (also PR from skyborla)

0.6

• mclogit now complains about (i.e. throws an error exception) when the random effects structure cannot be estimated, e.g. because random effects are constant within choice sets and therefore drop out by the formation of conditional logits.

0.5.1

• Fixed some bugs in predict models for mclogit objects.
• Made sure that dummy coding is used for response factors even if they are ordinal

0.5.0

New features

• Support for random slopes in multinomial conditional logit models
• Support for random intercepts and random slopes in multinomial baseline logit models

0.4.4

• Fixed mclogit.fit() and predict.mclogit() to work better without covariates.
• Explicitely import package "methods"

0.4.3

• Import as from package "methods".
• Make sure nobs is defined in mclogit.fit.rePQL.
• Updated DESCRIPTION file: Maintainer email address changed and no "This package" at start of package descriptions.
• Fix display of number of observations
• Drop redundant coefficients