semPlotModel: SEM model representation

Description

Methods to read a SEM object and return a semPlotModel-class object.

Usage

# S3 method for default
semPlotModel(object, ...)
# S3 method for lm
semPlotModel(object, ...)
# S3 method for principal
semPlotModel(object, ...)
# S3 method for psych
semPlotModel(object, cut, ...)
# S3 method for princomp
semPlotModel(object, ...)
# S3 method for loadings
semPlotModel(object, ...)
# S3 method for factanal
semPlotModel(object, ...)

# S3 method for lisrel
semPlotModel(object, ...)

# S3 method for mplus.model
semPlotModel(object, mplusStd = c("std", "stdy", "stdyx"), ...)
# S3 method for sem
semPlotModel(object, ...)
# S3 method for msem
semPlotModel(object, ...)
# S3 method for msemObjectiveML
semPlotModel(object, ...)
# S3 method for efaList
semPlotModel(object, which, ...)
# S3 method for psem
semPlotModel(object, ...)
# S3 method for wls
semPlotModel(object, ...)
# S3 method for seminr_model
semPlotModel(object, ...)
# S3 method for cSEMResults
semPlotModel(object, ...)
semPlotModel_Amos(object)
semPlotModel_Onyx(object)
semPlotModel_lavaanModel(object, ...)
semPlotModel_psychonetrics(object, delta = c("variance", "ignore", "show"), ...)
semPlotModel_lavaan_mi(object, ...)

Value

A "semPlotModel" object. See link{semPlotModel-class}

Arguments

object: An object contaning the result of a SEM or GLM analysis, or a string contaning the file path to the output file of a SEM program. Or a Lavaan model.
mplusStd: What standardization to use in Mplus models?
cut: Loadings with absolute value below this cutoff are omitted when importing psych fa/omega objects. Defaults to 0 (keep all).
which: Which solution to import from a lavaan efa() result (an efaList): a numeric index or a name such as "nf2". Defaults to the last (highest number of factors).
model: The original sem model (used in cvregsem)
delta: How to display the Delta scaling parameters of GGM-parameterized blocks in psychonetrics models. "variance" (default) replaces them by the implied variances (the diagonal of \(\Delta (I - \Omega)^{-1} \Delta\)), which are directly interpretable; "ignore" omits them; "show" displays the raw scaling parameters as self-loops.
...: Arguments sent to 'lisrelModel', not used in other methods.

Author

Sacha Epskamp <mail@sachaepskamp.com>

Details

A detailed overview of which packages are supported and what is supported for each of them will soon be on my website.

psychonetrics models (semPlotModel_psychonetrics, dispatched automatically by semPlotModel) are supported for the lvm and varcov frameworks, including all subtypes of their (co)variance structures: "cov", "ggm" (displayed as undirected network edges), "chol" and "prec" (both displayed as implied covariances). The development version of psychonetrics (remotes::install_github("SachaEpskamp/psychonetrics")) is recommended.

Pooled multiple-imputation models from the lavaan.mi package (semPlotModel_lavaan_mi, dispatched automatically) are displayed with Rubin's-rules pooled estimates and pooled standardized solutions; observed and implied covariance matrices are not defined for pooled fits, so semCors is unavailable for them.

PLS-SEM models from seminr (estimate_pls()) and cSEM (csem()) are supported: reflective constructs display directed loadings, composite (mode B / formative) constructs display weight arrows into the construct; PLS estimates are standardized. metaSEM stage-2 models (tssem2()/wls()) are supported through metaSEM's meta2semPlot(); stage-1 pooling objects give an informative error. umx models (umxRAM() returns an MxRAMModel) are supported via the OpenMx importer. psych's fa() and omega() outputs are supported (omega as a bifactor structure; pair it with semPaths(..., bifactor = 'g', layout = 'tree2')); principal() was already supported.