ESEM-within-CFA syntax composer for lavaan

Usage

syntax_composer(efa_object, referents, only_fix_crossloadings = TRUE)

Arguments

efa_object: A psych::fa() object with the results of a exploratory factor analysis. The factor loadings from this analysis will be set as starting values or fixed values in the lavaan syntax created.
referents: A named list in the form list(factor1 = "referent1", etc). Each entry identifies the referent for one of the factors. The order of this vector must be the same order the factors are ordered in the efa_object.

Value

A character vector with a lavaan syntax for the ESEM model.

Details

ESEM-within-CFA models have latent variables estimations with all cross-loadings. To improve fit, starting values from a previous exploratory analysis are used. To ensure model identification, a referent is chosen for each latent variable and all their cross-loadings are set as fixed values. This function "composes" such a lavaan model syntax from an EFA object. This model can subsequently be used in lavaan with lavaan::cfa() or lavaan::sem(). The best way to check the output is with writeLines().

It is very important to make sure the order of the referents vector is the same as the factor order in the factor loadings matrix in efa_object. Otherwise, the names of the latent variables may be wrong in the syntax. See also the vignette esem-within-cfa.

Examples

# use Holzinger and Swineford (1939) dataset in lavaan package
hw_data <- lavaan::HolzingerSwineford1939
hw_data <- hw_data[,c(7:15)]

#make exploratory analysis with geomin rotation
geomin_efa <- esem_efa(hw_data,3)
referents_list <- list(textual = "x5", visual = "x3", speed = "x7")
model_syntax <- syntax_composer(geomin_efa, referents_list)
writeLines(model_syntax)
#> textual =~ start(0.193)*x1+
#> start(0.042)*x2+
#> -0.066*x3+
#> start(0.845)*x4+
#> start(0.884)*x5+
#> start(0.804)*x6+
#> 0.035*x7+
#> start(-0.044)*x8+
#> start(0.024)*x9 
#> 
#> visual =~ start(0.595)*x1+
#> start(0.509)*x2+
#> start(0.688)*x3+
#> start(0.02)*x4+
#> -0.061*x5+
#> start(0.083)*x6+
#> -0.137*x7+
#> start(0.137)*x8+
#> start(0.391)*x9 
#> 
#> speed =~ start(0.029)*x1+
#> start(-0.124)*x2+
#> 0.016*x3+
#> start(0.009)*x4+
#> 0.009*x5+
#> start(-0.011)*x6+
#> start(0.731)*x7+
#> start(0.688)*x8+
#> start(0.455)*x9 
#> 

# esem-within-cfa
esem_w_cfa <- lavaan::cfa(model_syntax, data = hw_data, std.lv = TRUE)
summary(esem_w_cfa)
#> Length  Class   Mode 
#>      1 lavaan     S4