LoginSign Up
← Back to all sparks
S

semmcci

ANALYTICS
Velocity0.0

R package semmcci by jeksterslab — release notes from GitHub.

github.com/jeksterslab/semmcci ↗

Monte Carlo confidence intervals for SEM, now mostly reacting to upstream deprecations

structural-equation-modelingmonte-carloconfidence-intervalsr-packagelavaan
◆Current state
semmcci generates Monte Carlo confidence intervals for structural equation model parameters, working alongside lavaan. Its four-year release history is a run of patch versions from the jeksterslab account, each adding a function or adjusting method detail. The recent ones are quieter still: the latest addresses a lavaan::getCov() deprecation in tests, and the one before it is described only as minor method edits.
◆Where it's heading
The functional build-out finished some time ago. MCGeneric() in 1.1.3 and Func()/MCFunc() in 1.1.4 opened the package to user-defined functions of parameters, which is the natural end point for a Monte Carlo interval tool — once arbitrary functions are supported, there is little left to add. Since then releases have tracked lavaan's changes rather than semmcci's own direction, and the gap between them has stretched from months to over a year.
◆Prediction
Expect the next release to be triggered by another lavaan deprecation rather than by new capability.

◆Recent moves

  1. 2mo ago

    lavaan getCov() deprecation handled in tests

    Addresses a lavaan::getCov() deprecation in the test suite. Test-side only, with nothing reaching what the package computes.

    View source ↗
  2. 10mo ago

    Minor method edits

    Minor edits to methods, with no detail in the release note.

    View source ↗
  3. 2y ago

    User-defined parameter functions via Func() and MCFunc()

    Func() and MCFunc() arrive, letting users define arbitrary functions of model parameters and get Monte Carlo intervals for them. Together with MCGeneric() this closes out the package's flexibility story.

    View source ↗
  4. 2y ago

    MCGeneric() opens up arbitrary parameter targets

    MCGeneric() extends interval generation beyond the built-in parameter set, alongside adjustments to seed handling for reproducibility.

    View source ↗
  5. 3y ago

    Multiple-imputation support via MCMI()

    MCMI() adds multiple-imputation support, and the Behavior Research Methods paper is registered as the package citation.

    View source ↗
  6. 3y ago

    Data generation internals refactored

    Internal refactoring of data generation functions, with no user-visible effect.

    View source ↗