LoginSign Up
← Back to all sparks
L

lavaanExtra

ANALYTICS
Velocity0.0

R package lavaanExtra by rempsyc — release notes from GitHub.

github.com/rempsyc/lavaanExtra ↗

SEM reporting helpers converging on APA output, one CRAN resubmission at a time.

r-packagestructural-equation-modelinglavaanapa-reportingcran
◆Current state
lavaanExtra provides shorthand syntax and formatted output around lavaan structural equation models - write_lavaan() to build model strings, and nice_* functions for fit tables, plots, and modification indices. Three of the six visible releases exist only to satisfy CRAN resubmission: a unicode problem, a dependency version check, tests running without suggested packages. The substance sits in 0.1.5, 0.1.8, and 0.1.9.
◆Where it's heading
The package generalises its own vocabulary as it goes: lavaan_ind() became lavaan_defined() once it turned out to extract any user-defined parameter, and lavaan_cov() was split so lavaan_cor() covers actual correlations. Methodological positions are taken alongside the API - dropping the estimate argument from lavaan_reg() to force reporting both standardized and unstandardized values, and updating the RMSEA benchmark to Schreiber (2017). Rémi Thériault maintains it next to rempsyc, which formats output to match. Note that 0.1.5 restates the whole 0.1.4.x development series in one body.
◆Prediction
The pattern points to another nice_* helper aimed at a reporting step that currently needs hand formatting, arriving with the usual CRAN resubmission behind it.

◆Recent moves

  1. 2y ago

    CRAN resubmission for a unicode problem

    A version bump six days after 0.1.9 purely to clear a CRAN encoding complaint. The version number moves; the package does not.

    View source ↗
  2. 2y ago

    lavaan_ind renamed to lavaan_defined; thresholds supported

    The rename is the point: the function was never specific to indirect effects, and lavaan_defined() now names what it actually extracts. Dropping the estimate argument from lavaan_reg() is the same instinct applied to reporting practice rather than naming.

    View source ↗
  3. 2y ago

    nice_modindices flags redundant items

    Pairs modification indices with item labels and a similarity score, so a suggested path can be read as item redundancy rather than a model fix. A reporting helper that carries an interpretive opinion.

    View source ↗
  4. 3y ago

    Suggested dependency versions checked correctly

    One of three CRAN resubmissions in the window, this time over how suggested packages are version-checked.

    View source ↗
  5. 3y ago

    Tests run without suggested dependencies

    A CRAN policy fix two days after 0.1.5. The cluster of resubmissions in mid-2023 says more about CRAN's checks than about the package.

    View source ↗
  6. 3y ago

    Fit benchmarks updated and correlations split from covariances

    A cumulative note that restates five development versions at once, so its length overstates what shipped on this date. The substance is methodological: the RMSEA benchmark moves to Schreiber (2017), lavaan_cor() splits off from lavaan_cov(), and vignettes leave the package to meet CRAN size limits.

    View source ↗