- CRAN resubmission to fix unicode problem
- New release + CRAN
- New paper in JOSS!
Citation updated accordingly.
?lavaanExtra now works as expected.write_lavaan(): accepts a new argument,
threshold, represented by the “|” operator.lavaan_ind() renames to lavaan_defined()
because it was not specific to indirect effects but actually extracts
any user-defined parameters such as total effects and moderated slopes
using the := operator.lavaan_reg() gets rid of the estimate argument, to
comply with best practices of reporting both unstandardized and
standardized parameters with their respective confidence intervals.- new function:
lavaan_extract(), which takes a specific
operator and extracts relevant information (now used internally for
lavaan_reg(), lavaan_defined(),
lavaan_cov())
- CRAN resubmission
- New function:
nice_modindices, which outputs
modification indices along item labels and a similarity score between
the left-hand side and right-hand side items. Useful to assess item
redundancy when considering modification indices.
- CRAN resubmission: suggested dependencies now check for the correct
version with
rlang::check_installed().
- CRAN resubmission: tests now run even without suggested
dependencies.
nice_tidySEM: new argument: reduce_items,
to change the size of the nodes (boxes) of the items defining the latent
variables (which often require less emphasis).
nice_tidySEM: now omit the leading zero when using
standardized coefficients (for APA style).nice_fit:
- Update fit indices recommendations from Schreiber et al. (2006) to
Schreiber (2017). Only benchmark that has changed is RMSEA, moving to
< .05 [.00, .08] from
< .06-.08 [.00, .10].
- Make table shorter by combining the RMSEA column and its confidence
interval and shortening the AIC and BIC interpretations.
lavaan_cov: renamed r column to covariance since in
some cases standardized residual variances were not real
correlations.- New function:
lavaan_cor, which is the same as
lavaan_cov but only for actual correlations
lavaan_reg, lavaan_ind, and
lavaan_cov: new argument estimate which can be
either “B” to obtain standardized estimates and corresponding p values
and confidence interval, or “b” for unstandardized values.nice_lavaanPlot: stars now appear not only for
regression per default, but also for latent variables and covariances
c("regress", "latent", "covs").
nice_tidySEM: corrected a bug whereas if the layout
contained columns named c(“IV”, “M”, “DV”), it would remove any extra
columns (such as items), explicitly ignoring part of the layout
provided.nice_lavaanPlot: now default to black path lines
instead of gray.
- In examples and internal tests, now check for package availability
to satisfy CRAN requirements.
- Remove vignettes from package, to satisfy CRAN package size
requirements (they are still available on the website however).
nice_fit: better error message when not providing a
lavaan object.tmvnsim package now required for nice_tidySEM as it
seems necessary to use tidySEM’s new version.
nice_tidySEM: New argument to hide covariances:
hide_cov (defaults to FALSE).
- Breaking changes:
- We remove the copied/reexported
save_as_docx and the
like in favour of explicitly calling
flextable::save_as_docx. This way we are not creating
namespace conflicts for these functions, we are consistent with the new
approach in rempsyc, and it also gives credit to the
flextable package, as this is the powerhouse that produces
the tables under the hood.
nice_fit now requires a list of models (or a single
model), so it will not accept models loosely provided in the function as
different arguments. This is because we get rid of the dot-dot-dot
... argument in favour of supporting list objects.
nice_fit now has an added footnote reference to
Schreiber et al. (2006) for fit indices references when using
nice_table = TRUE.
nice_fit: gains a model.labels argument to
customize the model names in the table.
lavaan_ind gains an argument,
underscores_to_arrows, to replace underscores by arrows for
the indirect effect column, so that indirect effects are more quickly
visually interpretable.nice_tidySEM: More meaningful error message when
providing unmatching label names.nice_fit look improved when using
nice_table = TRUE
rempsyc package dependency is now on CRAN, so we can
rely on it normally now (yeah!).- Adding test coverage (100% so far!).
- Preparing for CRAN.
nice_tidySEM: Gains the label_location and
... arguments.
nice_tidySEM: Corrected bug when not providing a
structure/matrix layout.
- Added new function:
nice_tidySEM, for a quick and
decent-looking tidySEM plot with sensical (but
customizable) defaults.
write_lavaan: Exciting news! Automatic indirect effects
now seemingly support any combination of IV, moderator, and DV!
write_lavaan: Automatic indirect effects now support
another scenario: two mediators and three IVs!
- Added new function:
nice_lavaanPlot, for a quick and
decent-looking lavaanPlot with sensical (but customizable)
defaults.
write_lavaan: Automatic indirect effects now support
another scenario: two mediators and two IVs!
write_lavaan: Fixed bug with automatic indirect effects
when more than 1 mediator making forbidden paths (two mediators tested
successfully)
cfa_fit_plot:
- Adds ellipsis to pass any desired argument to
lavaan::cfa, and thus removes the default
missing = "fiml" which was causing a bug with
estimator = DWLS.
- Changes the default from saving to PDF to opening in RStudio Viewer,
and adds an additional
save.as.pdf argument to save as
PDF.
- Now imports
flextable::save_as_docx (and
save_as_html, save_as_image,
save_as_pptx…) to support saving the nice_fit
tables to Word without having to explicitly load flextable
or rempsyc.
- Added package logo!
- Added new function:
cfa_fit_plot to simultaneously
print, save, and plot the CFA results.
- Added new arguments/symbols:
constraint.equal
(==), constraint.smaller (<),
constraint.larger (>), custom
(accepts a single, lavaan-like string)
- Added automatic indirect effects