| NEWS | R Documentation |
a few minor internal improvements (no user-visible changes)
the margEff() method for objects of class censReg
now has an additional argument vcovLogSigma,
which defaults to TRUE;
this change makes the margEff() method consistent
with the coef() and vcov() methods
but it changes the default way of calculating marginal effects
compared to the previous version
added argument subset to function censReg()
added argument vcov to the margeff() method
for objects of class censReg.
made the tests of this package compatible with version 1.4-0
of the maxLik package
censReg() can now estimate models that only have an intercept
and no (other) explanatory variables
made the output of the test scripts much more robust to differences in numerical inaccuracies
some minor internal improvements and modifications, e.g. to make the 'censReg' package compatible with future versions of the 'plm' package
added argument 'xValues' to the margEff() method so that the user can specify
the values of the explanatory variables, at wich the marginal effects are
calculated
censReg() now stops with an error message if there are no uncensored
observations or there are no censored observations
fixed an internal problem so that this package can be used with R version 3.1
fixed a few minor internal problems that were indicated by "R CMD check" with R version 3.1
corrected bug in the calculation of the covariance matrix of the marginal effects: hence, all statistical tests for the marginal effects were wrong in previous versions of the censReg package
added print() method for objects that are returned by the margEff() method
the calculation of the gradients of the log-likelihood function for cross-sectional data is numerically more stable
added nobs() method that returns the (total) number of observations
added margEff() method that computes the marginal effects of the explanatory
variables on the expected value of the dependent variable evaluated at the mean
values of the explanatory variables. This functionality is currently not
available for panel data models.
censReg() has an argument "logLikOnly" now: if this argument is set to TRUE,
the model is not estimated but the log-likelihood contributions of all
observations/individuals and the corresponding gradients calculated at the
parameters specified by argument "start" are returned
made the calculation of the log-likelihood value of random effects panel data models with censored dependent variables and the calculation of the corresponsing gradient(s) numerically more robust and stable by calculating and storing the *logarithms* rather than the actual numbers
fixed bug that made it impossible to specify starting values for the estimation of censored random-effects panel data models (many thanks to Igors Lahanciks for reporting this bug!)
fixed bug that occured when a factor had "empty" levels
(many thanks to Ulrich Morawetz for reporting this bug!)
censReg() now additionally returns the model terms and
the degrees of freedom of the residuals
added a model.frame() method for objects of class "censReg"
added a formula() method for objects of class "censReg"
added extractAIC() method for objects of class "censReg"
added estfun() method for objects of class "censReg"
that has an argument "includeSigma" (defaults to TRUE),
which determines whether the derivative with respect to the
logarithm of the variance parameter should be included
initial public release of the censReg package