Changes in Version 0.9-40
* bptest() now optionally also accepts a "weights" argument and processes
weighted "lm" objects correctly (suggested by Robert Gulotty).
* The degrees of freedom (df) in the default coeftest() method can now also
be a vector of the same length as the coefficients.
Changes in Version 0.9-39
o Small improvements in handling of "df" attribute in coeftest() methods.
o In case update() is called within the default waldtest() method, it tries
harder to evaluate this in the calling environment (reported by Vincent
Arel-Bundock). This is still not guaranteed to find the right data but
should be more robust than the previous implementation. To avoid any such
issues it is recommended to call waldtest(object1, object2) and fitted
model objects (and not just update formulas).
Changes in Version 0.9-38
o coeftest.default() now includes attributes for "nobs" and "logLik" provided
that nobs() and logLik() extractors are available. This is in order to
facilitate model summaries a la broom::glance for "coeftest" objects
(suggested by Grant McDermott). The attributes can be extracted again
from the coeftest object with nobs() and logLik(), respectively.
o Optionally, the entire original model object x is saved as an attribute
"object" in coeftest(x, save = TRUE). Like nobs/logLik above, this should
also facilitate model summaries.
o waldtest.default() is stricter about detecting non-nestedness of models,
e.g., when under the alternative variables are added but the intercept
dropped.
o waldtest.lm() now uses . ~ 0 as the default reference model if waldtest(object)
tests an object without intercept (suggested by Kevin Tappe).
Changes in Version 0.9-37
o coeftest() gained a "df" attribute facilitating subsequent processing
of its output, e.g., for computing the corresponding confidence intervals.
Suggested by Alex Hayes in https://github.com/tidymodels/broom/issues/663.
o Based on the new "df" attribute of "coeftest" objects, a method for
confint() is added. confint(coeftest(object, ...)) should match the output
of coefci(object, ...).
o Based on the new "df" attribute of "coeftest" objects, a method for
df.residual() is added. df.residual(coeftest(object, ...)) returns NULL
if a normal (rather than t) approximation was used in
coeftest(object, ...) even if df.residual(object) returned something
different.
Changes in Version 0.9-36
o resettest() gained a vcov and ... argument that - if present - are
passed to waldtest() to carry out the F test. This enables the
usage of robust sandwich covariances etc.
o Various documentation improvements suggested by Kevin Tappe.
o Added native routine registration for pan.f.
Changes in Version 0.9-35
o Added a new function coefci() to accompany coeftest() for computing
the Wald confidence intervals for coefficients, e.g., based on
sandwich covariances.
o The default coeftest() method -- and also coefci() -- now pass the
... arguments on to the vcov.() function argument (if any).
o Fixed bug in bptest() for degrees-of-freedom calculation when
some regressors are aliased/collinear. Reported by Amrei Luise
Stammann.
o Various documentation improvements suggested by Kevin Tappe.
Changes in Version 0.9-34
o Registered all default methods as S3 methods (in addition to fully
exporting them).
o Reference output updated for recent versions of R.
Changes in Version 0.9-33
o Extended license to GPL-2 or GPL-3.
o To make resettest() results numerically more reliable (especially
for type = "fitted" when the fitted mean is large), the response
is internally scaled.
o The default coeftest() method checks whether coef() and vcov()
output has unique names (which they may not have for some model
classes, e.g., sampleSelection::selection).
Changes in Version 0.9-32
o Small improvements in residual degrees of freedom computation
within default waldtest() method.
o Updated Imports/Suggests to conform with current R CMD check.
Changes in Version 0.9-31
o Included Farebrother/Cummins as "aut" in DESCRIPTION for pan.f
Fortran code.
o Made starting values in the auxiliary regression of bgtest()
optional. By default still 0, but could also be NA.
o Avoid spurios errors about missing nobs() methods in lrtest().
Changes in Version 0.9-30
o coeftest(), lrtest(), and waldtest() can now also be used for
S4 objects provided that methods to the corresponding S4 generics
from stats4 are supplied.
o Added output of examples and vignettes as .Rout.save for
R CMD check.
o Moved vignette to "vignettes" directory in source package.
Changes in Version 0.9-29
o dwtest() now catches weighted regressions and throws an error
because weighted regressions are not supported (yet).
o bgtest() now returns an object of class "bgtest" (inheriting from
"htest"). This has coef(), vcov() and df.residual() methods
- and thus enable calling coeftest() - to show the results of the
auxiliary regression.
Changes in Version 0.9-28
o Fixed typo in print output of dwtest() for two-sided alternatives.
o Fixed problem in coeftest() method for "breakpointsfull" objects
(from strucchange). The vcov() method was not called correctly,
resulting in an error.
Changes in Version 0.9-27
o Modified grangertest() default method to workaround a bug
in as.zoo().
Changes in Version 0.9-26
o added coeftest() method for "mlm" objects
(needed for matching coef() output to vcov() method)
Changes in Version 0.9-25
o enhanced documentation for new Rd parser.
Changes in Version 0.9-24
o Added new specification test: PE test for linear vs.
log-linear specification in linear regressions is
now available as petest().
Changes in Version 0.9-23
o enhanced documentation for new Rd parser.
Changes in Version 0.9-22
o included lmtest.bib in inst/doc for vignette.
o removed \itemize in .Rd files for new R-devel.
Changes in Version 0.9-21
o corrected degrees of freedom for bgtest(..., type = "F")
(bug spotted by Oscar Becerra).
Changes in Version 0.9-20
o fixed error in Mandible data. Mandible[158,2] was 34 but
should really be 37 (see Table 1 in Royston & Altman 1994).
Spotted by Christian Ritz.
Changes in Version 0.9-19
o changed the default behaviour of waldtest.default():
Chisq test (instead of F test) is now used by default
which is more sensible for most models (with linear
models as a notable exception).
o added waldtest.lm() method that switches the default
test back to F test (instead of Chisq test).
Changes in Version 0.9-18
o forgot to export print() method for "coeftest" objects
Changes in Version 0.9-17
o added NAMESPACE
o improved dependency declaration in DESCRIPTION
Changes in Version 0.9-16
o updated all .rda data sets to RDX2
o enhanced sanity checking in dwtest()
Changes in Version 0.9-15
o new generic with flexible default method for lrtest(),
implementing asymptotic likelihood ratio tests. This works
if a logLik() method is provided and suitable methods for
updating models and checking consistency.
o simplified print output in coeftest()
Changes in Version 0.9-14
o now a default method for waldtest() is provided (replacing the
lm method). It works for "lm", "glm" and "survreg" objects.
Furtheremore, it is object-oriented and generally works if the
following methods are available: terms(), formula(), coef() (which
needs to be named, matching the names in terms() and vcov()),
vcov() (which can alternatively be user-supplied), df.residual(),
residuals() and update() (unless only fitted objects are supplied).
o the default coeftest() was enhanced, works now out of the box for
"mle" objects, a "survreg" method is added (needed because coef()
and vcov() do not necessarily match for "survreg" objects).
o the vcov argument was changed (almost) everywhere to vcov. to
avoid name clashes with the vcov() function.