Changelog

Stop calling tidySEM functions as default arguments of prepare_graph.lavaan(). This ensures that tidySEM works when its namespace is not attached.
Frank Gootjes contributed pseudo class functionality, new functions are append_class_draws() and pseudo_class()

Update maintainer email address
Fixed bug in plot_profiles() when covariances are free
Improved starting values for latent class analysis
Documentation incorrectly referred to data_mixture_ordinal, instead of data_mix_ordinal
Added legend to plot_probs()
Added np_ratio and np_local to table_fit() output for mixture models
Reimplement BLRT() to avoid slow OpenMx implementation
Add functions lr_test() and lr_lmr()

Fix Issue #45 , argument label was not passed correctly to get_nodes() and get_edges() in graph_sem() and prepare_graph()
Fix Issue #47, list of character vectors was not properly converted to an mxModel

Fix CRAN check errors related to OpenMx
Add mx_lca(), a dedicated function for estimating mixture models with ordinal/binary data that avoids convergence problems when using mx_mixture() with those data types
Add table_prob() to tabulate category probabilities for models with ordinal indicators
Add plot_prob() to plot category probabilities for models with ordinal indicators

Removed LazyData entry from DESCRIPTION (no data included in package)
Add methods for mplusObjects
Fixed bug where nodes from multilevel models were incorrectly merged, due to lack of within-level intercepts
Fixed bug where get_layout did not work for multilevel models because it used table_results() instead of get_edges()
Fixed bug in create_scales

Fixed bug where table_results() was broken by renaming ‘label’ column to ‘mplus_label’

Fixed several bugs in plotting, caused by multiple columns called “label” generated by mplus and lavaan, which overlap with the “label” column used for plotting. Renamed other “label” columns to “mplus_label” and “lavaan_label”
Fixed bug where arguments were not passed from prepare_graph() and graph_sem() to get_nodes() and get_edges(), e.g., digits was not passed on
Fixed bug where linetype did not work when using a mix of numeric (e.g., 1) and character (e.g., “dashed”) values
Node labels can now be customized just like edge labels
Added convenience functions to conditionally customize graphs based on if_edit(), e.g.: sig_colour(plot, “green”)
Improved functionality of edit_graph() methods
Added wrappers for edit_graph(): edit_edges() and edit_nodes()
graph_sem(fix_coord = TRUE) now works for multi-group models

Fixed bug in direction of arrow for curved paths
Support argument “color” in addition to “colour”
The “label” argument of get_edges() and get_nodes() now accepts an expression, that is evaluated in the context of a data.frame generated by table_results(). This allows users to specify compound labels.

Changed arguments for graph_sem.lavaan and graph_sem.mplus.model; these gain arguments “label”, “edges”, “layout”, and “nodes”, which apply get_edges(), get_layout(), and get_nodes() to the model argument, respectively, passing on the “label” argument to get_edges() and get_nodes().
The default label has been changed to “est_sig” throughout the package. Some functions used “est_sig_std” instead.
Minor bugfixes
Fix bug where > 26 nodes resulted in wrong plot

create_scales() no longer writes .csv files
add_paths() now relies on lavaanify(lavParseModelString()), uses the lavaan parameter table, and accepts all arguments of lavaanify(), with default arguments same as sem() and cfa()
measurement() uses add_paths() for consistency
graph_sem.lavaan() and graph_sem.mplus.model() gain an argument label = "est_sig", so users can easily select custom labels.
New vignette about SEM plotting conventions.
graph_sem() adds empty labels for nodes without a mean; in version 0.1.1, such nodes were displayed without any node label.

tidySEM 0.2.7