Breaking Changes: Added the arguments
cross_add and cross_sets to
model_set() and get_add() to better handle
factor loadings. Some factor loadings are no longer added by default,
such as an indicator loading on two latent factors and one of the
factors regresses on another. These cross-loadings are not meaningful in
some models and may lead to nonconvergence. To reproduce results in
previous versions of modelbpp, users may need to manually
add these loadings using cross_add and
cross_sets. (0.1.6.3)
Breaking Changes: Added the argument
loadings_to_exclude_from_drop to model_set()
(and loadings_to_exclude to get_add()) to
better handle factor loadings. Some factor loadings will no longer be
considered to be dropped by default, such as an indicator that loads on
only one latent factor. Cross-loadings will still be considered. To
reproduce results in previous versions of modelbpp, users
may need to set loadings_to_exclude_from_drop to
"none". (0.1.6.3)
Added a few more tests (0.1.6.2)
Updated gen_models() with the new arguments added to
model_set(). (0.1.6.4)
add_list_duplicate_cov(), which makes
must_not_add failed to exclude some covariances. It should
work now. (0.1.6.1)model_set_combined() for computing BPPs for
models from two or more calls to model_set(). (0.1.5.3,
0.1.5.4)The default of more_fit_measures of the print method
of model_set-class object was changed to
c("cfi", "rmsea", "srmr). (0.1.5.1)
Changed the vignettes to precomputed Rmarkdown files. (0.1.5.2)
Added the argument exclude_xy_cov and
exclude_feedback to get_add() and
model_set(), for excluding paths that create feedback
loops, and covariances involving a predictor and an outcome variable
(including those linked by indirect paths). Default values has been
changed to TRUE since 0.1.3.5. To reproduce results from
previous version, set them to FALSE. (0.1.3.2,
0.1.3.5)
Added min_bpp_labelled to
model_graph(), to hide the labels of models with small
BPPs. (0.1.3.5)
Added the argument drop_equivalent_models, to
model_set(). If TRUE, the default, the models
fitted will be checked for equivalence. If two or more more models are
equivalent, only one of them will be retained. The groups of equivalent
models identified, and the models dropped, will be printed by the print
method. (0.1.3.9)
Added measurement_invariance_models(), for
generating metric and scalar invariance models and their partial
invariance versions. (0.1.3.10 - 0.1.3.11)
Because it is very likely that users would like to see come fit
measures along with BPPs, the default of more_fit_measures
of the print method of model_set-class object changed to
c("cfi", "rmsea"). (0.1.3.7)
Revised fit_many() to support multigroup models.
(0.1.3.8)
A progress bar can be displayed when model_set() is
identifying nested models. (0.1.3.13)
Shortened BIC Posterior Probability to
BPP in some sections of the printout of
print.model_set(). (0.1.3.14)
Cumulative BPPs no longer displayed by default in
print.model_set(). Print them by setting
cumulative_bpp to TRUE. (0.1.3.15)
Update an internal function to handle nonconvergence in checking nested relation. Only affect the graphs and only happen in some rare cases. (0.1.3.16)
The must_not_add argument should work now for some
parameters that may not be recognized as interchangeable.
(0.1.3.1)
Fixed a bug in must_not_drop and
must_drop of get_drop(). They should work
properly now. (0.1.3.5)
Fixed a bug in model_graph(). Short names should now
be properly constructed. (0.1.3.3)
Fixed some bugs in print.model_set() about the
printing of additional fit measures. (0.1.3.6, 0.1.3.7)
Fixed a bug in checking whether two models are equivalent. (0.1.3.12)
model_set()model_set() to work with user-supplied models.
These models are supplied as parameter tables through the argument
partables. (0.1.2.7)print-method of
model_set-class objects. Users can set the prior
probabilities of one or more models of their choice. (0.1.2.7)c-method for partables-class and
model_set-class objects. For the ease of adding user models
when calling model_set(). (0.1.2.7)lavaan-class objects) to model_set() through
the argument sem_out. (0.1.2.17)print method of model_set() supports
printing additional fit measures available from
lavaan::fitMeasures(). Check the argument
more_fit_measures. (0.1.2.27)print method of model_set() support
printing short model names, which can be used to interpret the output of
model_graph(). (0.1.2.29, 0.1.2.30)model_graph() to determine nested relation
using the method by Bentler and Satorra (2010). This can be done only if
fixed.x is set to FALSE. (0.1.2.10,
0.1.2.19)model_graph() if nested
relations need to be determined. (0.1.2.28)model_graph()model_graph() to plot user-supplied models.
(0.1.2.7)model_graph() with new options. If
drop_redundant_direct_paths is TRUE (default),
redundant direct paths will be removed.model_graph(). (0.1.2.20)model_graph() to label
arrows by model df differences (see
label_arrow_by_df), and weight arrow widths by model
df differences (see weight_arrows_by_df).
(0.1.2.20)partables-class
object. (0.1.2.8)partables-class
objects. (0.1.2.9, 0.1.2.12)model_graph() to use short names in
the graph, if they are created and stored by model_set().
(0.1.2.29, 0.1.2.30)lavaan::modificationIndices()
about equality constraints. (0.1.2.5)unique_models() to handle user-supplied models.
(0.1.2.6)model_set(). (0.1.2.11)print method of
partables. (0.1.2.13)model_set() will check whether the sum of user-supplied
prior probabilities is less than 1. (0.1.2.14)print method of model_set
objects to print original model dfs. (0.1.2.15)fit_many(), can set the model with
whichfit_many() will compute model df difference.
(0.1.2.16)model_set-class object to
print models that failed the past.check of lavaan.
(0.1.2.32)model_set() if no
paths are dropped or no paths are added. (0.1.2.3)gen_models()’s
argument, output.tinytest for tests. (0.1.0.9004 - 0.1.0.9006)