Fixes
* Fixed anova plot labelling issue with two-way interaction
* Fixed model_summary
print lm model summary when glm model
is specified
* Fixed model summary
cannot handle aov
models
* Fixed the check_factorstructure
function to import from
performance
instead of parameters
package
Major Feature
* Added support for ANOVA plots (with continuous variable as
moderator)
* Added support for polynomial plot (incl. curvilinear plots)
* Added support for Cronbach alpha computation (useful to combine with
descriptive_table
)
Minor Feature
* Integrate two-way and three-way interaction plot into one function
Fixes
* Fixed the issue that some model assumption checks were not
printed
* Fixed the issue that compare_fit
function not able to for
comparing lm
models
Fixes
* Fixed control variables must be numeric variables in interaction plots
(i.e., added support for factor variables) * Simple slope no longer need
to pass in the interaction_terms
and data
arguments
* Support multilevel modeling again after fixes introduced in
insight
package.
Fixes
* Drop support for multilevel modeling temporarily due to
insight
package recent non-backward compatible changes
Fixes
* Fixed bugs that measurement invariance does not have row name.
Major Feature
* Added support reliability
measure summary
* Added support mediation models
* Added support generalized linear regression (glm
and
glmer
without plot)
Minor Feature
* cfa_summary
support path diagram
* efa_summary
rewrite using functions from
parameters
and support post-hoc CFA test
* cfa_summary
support factor loading is hidden for same
latent factor (only when group = NULL
)
* cor_test
and descriptive_table
support
rich-text formatted table output * model_summary
rewrite
using parameters::model_parameters
* integrate summary with plot for lm_summary
to
integrated_model_summary
* cor_test
re-write
using the correlation package, so it supports more methods and robust
standard errors
* quite
and streamline
support in all models
that print output
* Give instruction on how to use R Markdown (see
knit_to_Rmd
)
Major Feature
* Added support linear regression
* Added support exploratory
factor analysis
* Complete overhaul to produce rich-text formatted
summary output
Minor Feature
* measurement_invariance
support multiple-factor model with
tidyselect syntax
* model_summary_with_plot
support
outlier detection
* Changed data from EWCS_2015_shorten to popular
(a data-set that is easier to understand)
* Added a new function
that allow convert HTML to PDF function for knitting Rmd
*
model_performance
support a wider array of model
performance measure
* cfa_summary
and
measurement_invariance
support checking goodness of fit
Fixes
* Critical bug fix for model_summary_with_plot
. You can no
request simple_slope
and check_assumption
correctly.
* Critical bug fix that cor_test
is not
exported
* remove some packages from import and switch to
requireNamespace()
* added fallback for normality
check
Major Feature
* lme_model
, model_summary_with_plot
support
tidyselect syntax
* cfa_summary
support multi-factor
CFA with tidyselect syntax
Minor Feature
* Added assumption_plot
to visually inspect assumptions for
mixed effect models in model_summary_with_plot
*
two_way_interaction_plot
,
three_way_interaction_plot
only require the model object to
produce the plot.
* lme_model
,
model_summary_with_plot
support using lme4
package.
* model_summary_with_plot
lme_model
support passing explicit model
*
compare_fit
support more model comparison (e.g., lme, glme)
Fixes
* This current version build pass CMD check
*
measurement_invariance
stop using
semTools::compareFit
. Added a self-created
compare_fit
function for the package
* Remove
papaja::apa_theme()
dependency.
* Use
.data
from rlang
for mutate
function
* model_summary_with_plot
always return list
and changed to logical (set to T to return result_list)
*
model_summary_with_plot
return a named list object
New Feature
* descriptive_table
support wider array of descriptive
indicator (e.g., median, range) and missing & non_missing values
count
Fixes
* Fixed the cor_test
bug that the function return a
correlation matrix with blank cells if the correlation is too high
between the two items (rounded to 1).
* Add a
data_check
function that warns the users if non-numeric
variables are coerced into numeric.