Changes in 0.5-4

  * citysize (new data set): city-size paired-comparison task
  * recogROC.R, recogROC.Rd: first column contains percentage of target (old),
    not new, items (reported by Katharina Naumann)
  * *.Rd: add DOIs to all references
  * NAMESPACE: properly import from graphics

Changes in 0.5-3

  * anova.mpt: now works with stats::print.anova()
  * mpt: small improvements for mpt(..., method = "EM")

Changes in 0.5-2

  * BUG FIX: mpt(..., method = "EM") could fail when a symbol in the model was
    also used as an object name in the work space
  * mpt: newline and other space characters in model specifications are now
    also allowed for the EM fitting method (not only for BFGS)
  * logLik.mpt: add multinomial constant
  * NAMESPACE: import graphics, utils

Changes in 0.5-1

  * mpt: the start argument expects starting values between zero and one
    regardless of the fitting method
  * mptspec("prospec"): return specification of event-based prospective memory
    model (Smith & Bayen, 2004)
  * prospecMemory, recogROC (new data sets): prospective memory data (Smith &
    Bayen, 2004), recognition memory data (Broeder & Schuetz, 2009)

Changes in 0.5-0

  * User-visible changes: MPT models are now specified via the new mptspec()
    function. Its return value serves as an input to mpt() which fits the
    model to data. It allows for more flexible parameter restrictions than the
    earlier mechanism. It includes all the default MPT models (plus new ones)
    that were formerly returned by the (now defunct) mptmodel() function. See
    ?mptspec and ?mpt for details.
  * coef.mpt, vcov.mpt, confint.mpt: new logical argument logit to switch
    between logit-scale and probability-scale representation
  * mpt: new argument method = c("BFGS", "EM") to select the fitting method
  * vcov.mpt: new argument what = c("vcov", "fisher") to return either the
    covariance matrix or the Fisher information
  * predict.mpt (new function): predicted response frequencies or
    probabilities for a given MPT model and optionally new data
  * valence (new data set): word valence and source memory for vertical
    position

Changes in 0.4-1

  * BUG FIX: typo in vcov.mpt(), hence wrong standard errors (reported by
    Rainer Alexandrowicz and Bartosz Gula)
  * recogROC (new data set): recognition memory data (Broeder & Schuetz, 2009)
  * nobs.mpt (new function): extracts number of nonredundant response
    categories
  * Add citation file: useR! 2011 talk

Changes in 0.4-0

  * simulate.mpt (new function): simulate responses from a fitted mpt model
  * mptmodel (new function): return formulae for some prevalent MPT models
  * mpt.R: mpt() function now takes dataframe or vector input
  * NAMESPACE: new file

Changes in 0.3-0

  * BUG FIX: two frequencies in the proact data set were incorrect
  * vcov.mpt (new function): covariance matrix of mpt model parameters
  * summary.mpt (new function): summary method for mpt objects
  * mpt.Rd (examples): update
  * proact.Rd (examples): update

Changes in 0.2-0

  * BUG FIX: df in logLik.mpt() was incorrect by one
  * residuals.mpt (new function): residuals for MPT models
  * plot.mpt (new function): plots MPT residuals against fitted values
  * proact (new data set): recall frequencies for DaPolito's experiment on
    proactive inhibition
  * mptEM: new stepsize argument
  * mpt.Rd (value): update
  * mptEM.Rd (usage, value): update

Changes in 0.1-0

  * initial release on CRAN

