LCT {EGAnet}

R Documentation

Loadings Comparison Test

Description

An algorithm to identify whether data were generated from a factor or network model using factor and network loadings. The algorithm uses heuristics based on theory and simulation. These heuristics were then submitted to several deep learning neural networks with 240,000 samples per model with varying parameters.

Usage

LCT(
  data,
  n,
  iter = 100,
  dynamic = FALSE,
  dynamic.args = list(n.embed = 4, tau = 1, delta = 1, use.derivatives = 1)
)

Arguments

data	Matrix or data frame. A data frame with the variables to be used in the test or a correlation matrix. If the data used is a correlation matrix, the argument n will need to be specified
n	Integer. Sample size (if the data provided is a correlation matrix)
iter	Integer. Number of replicate samples to be drawn from a multivariate normal distribution (uses mvtnorm::mvrnorm). Defaults to 100
dynamic	Boolean. Is the dataset a time series where rows are time points and columns are variables? Defaults to FASLE.
dynamic.args	List. Arguments to be used in dynEGA. Defaults: n.embed Number of embeddings: 4 tau Lag: 1 delta Delta: 1 use.derivatives Derivatives: 1

Value

Returns a list containing:

empirical	Prediction of model based on empirical dataset only
bootstrap	Prediction of model based on means of the loadings across the bootstrap replicate samples
proportion	Proportions of models suggested across bootstraps

Author(s)

Hudson F. Golino <hfg9s at virginia.edu> and Alexander P. Christensen <alexpaulchristensen at gmail.com>

References

Christensen, A. P., & Golino, H. (2021). Factor or network model? Predictions from neural networks. Journal of Behavioral Data Science, 1(1), 85-126.

Examples

# Compute LCT
## Factor model
LCT(data = psychTools::bfi[,1:25])

## Not run: 
# Dynamic LCT
LCT(sim.dynEGA[sim.dynEGA$ID == 1,1:24], dynamic = TRUE)
## End(Not run)

---