MonteCarloSEM (version 2.0.0)

sim.skewed: Simulates Data Sets from a Structural Equation Model (SEM) with Normal or Non-Normal Distributions

Description

This function generates data sets based on a specified SEM. The simulated data are organized such that the first column represents case identifiers, while the subsequent columns contain the simulated item responses. For example, in a model with two factors and three items per factor, the column labels will follow the format: "ID, F1_x1, F1_x2, F1_x3, F2_x1, F2_x2, F2_x3". The number of rows corresponds to the sample number of the data. In addition to the generated data sets, two supplementary files are also saved: (1) "Model_Info.dat" — containing the factor correlation matrix, factor loading matrix, a vector indicating non-normal items, and the coefficients B, C, and D from Fleishman’s power method (where A = –C). (2) "Data_List.dat" — listing the names of all generated data files.

Usage

sim.skewed(
  nd = 10,
  ss = 100,
  fcors,
  loading,
  nonnormal = NULL,
  Fleishman = NULL,
  f.loc
)

Arguments

nd

An integer, the number of data sets to be generated.

ss

An integer, the sample size per data set (must be greater than 10).

fcors

The factor correlation matrix, which must be symmetric. For one-factor models, this should be "matrix(1,1,1)".

loading

The factor loading matrix. Columns correspond to factors, while non-zero rows specify the number of items associated with each factor.

nonnormal

A numeric vector of 0s and 1s indicating whether each variable should be generated as normal (0) or non-normal (1). If not specified, all variables are generated as normal.

Fleishman

A numeric vector containing the coefficients B, C, and D from Fleishman’s power method. Note that A = –C.

f.loc

File path indicating the directory where the generated data sets and auxiliary files will be saved.

Author

Fatih Orçan

Examples

fc<-fcors.value(nf=3, cors=c(1,.5,.6,.5,1,.4,.6,.4,1))
fl<-loading.value(nf=3, fl.loads=c(.5,.5,.5,0,0,0,0,0,0,0,0,.3,.3,.3,0,0,0,0,0,0,0,0,.4,.4))
ifN<-c(1,1,1,0,0,0,0,0)
fleis<-c(1.0174852, .190995, -.018577) # The coefficients for skewness = 1, kurtosis = 1

sim.skewed(nd=10, ss=100, fcors=fc,loading=fl, nonnormal = ifN, Fleishman = fleis, f.loc=tempdir())