powRICLPM 0.2.1
Minor improvements and fixes
- Fixed a bug in version 0.2.0, in which some of the column names of
the summary tables from
summary.powRICLPM.parameter() and
summary.powRICLM.condition() were misaligned due to a
missing column name.
- Fixed a bug in version 0.2.0, in which population values were not
correctly parsed when setting the
constraints = ...
argument of powRICLPM() to anything other than
constraints = "none" (the default). This resulted in
incorrect estimates of bias, mean square error (MSE), and
coverage (Cover). Estimates of power, minimum estimate
(Min), empirical standard deviation (SD),
average standard error (SE Avg) and accuracy were
unaffected and thus remained valid.
- Added additional documentation to
summary.powRICLPM()
about the interpretation of information in the summary table.
powRICLPM 0.2.0
Breaking changes
- The function
powRICLPM_Mplus has been superseded by
powRICLPM() where users can now set the argument
software = "Mplus" to use Mplus for their power analysis.
The old function has been removed in this version.
- The
bootstrap_reps argument of powRICLPM
is now deprecated. Instead, Monte Carlo standard errors are now based on
Morris et al. (2017), and available for all performance measures.
- The argument
alpha has been superseded by the
significance_criterion argument.
Minor improvements and
fixes
- The
reliability argument can now take in a vector of
reliabilities to simulate performance metrics under various levels of
item reliability.
- The
powRICLPM package now does not import the packages
dplyr and purrr anymore.
- The
cli package is now used for error handling.
- Slight speed and stability improvements when using
software = "lavaan".
- Analysis using
lavaan now relies on the default
lavTest settings. Setting lavTest to
tests = "none" lead to errors on some computers.
powRICLPM 0.1.1
Minor improvements and
fixes
- Now includes the
check_Phi() function, which users can
use to check if they have specified their Phi matrix of
lagged-effects as intended.
plot() now allows other performance measures, such as
bias, MSE, coverage, to be plotted through its y
argument.
- Mistakes in the model syntax of the estimation model when imposing
stationarity constraints (using
constraints = "stationarity") have now been corrected.
powRICLPM 0.1.0
New features
powRICLPM can now save the generated data sets by
specifying a path with the save_dat argument
Minor improvements and
fixes
- The
est_ME argument in powRICLPM has been
renamed estimate_ME.
- Internal model fitting using
lavaan now skips certain
checks to speed up the process.
- The
wSigma argument in powRICLPM has been
replaced with the within_cor argument. Now, only a
double denoting the correlation between the
within-components needs to specified rather than a correlation
matrix.
- By default,
powRICLPM now discards results from Monte
Carlo replications with inadmissible parameter results, unless bounded
estimation is used (bounds = TRUE).