Perform structural equation modeling (SEM) using lavaan
(Rosseel, 2012). Go to lavaan.org for tutorials on the model syntax. See also Kline (2015).
For additional reading, see https://osf.io/xkg3j/ for an introduction to SEM in JASP, by Burger & Tanis.
Throughout this help file some terms are used synonymously, these are: - latent variables and factors - observed variables, manifest variables and indicators
Here, users specify their model here in lavaan
syntax using the names of the variables in their data set (see https://lavaan.ugent.be/tutorial/sem.html). Multiple models can be specified and compared.
Assume factors uncorrelated: By default the factors are assumed to correlate if not specified differently in the model window. This checkbox changes that and factors are estimated to be orthogonal.
Fix exogenous covariates: If checked, the exogenous covariates are considered fixed variables and the means, variances and covariances of these variables are fixed to their sample values. If not checked, they are considered random, and the means, variances and covariances are free parameters.
Others:
Model test:
Browne residual based (NT): Browne's residual-based test statistic using normal theory is computed.
Information matrix: Matrix used to compute the standard errors
First order: The information matrix is based on the outer product of the casewise scores
Standard errors:
Bootstrap: Standard errors are computed from bootstrapped model fit objects.
Confidence intervals: CI width for the parameter estimates
f=~x2
Sample-size adjusted Bayesian Information Criterion (SSABIC): A version of BIC adjusted for sample size, with lower values indicating a better fit.
T-size fit indices: Equvalence testing: Testing for close fit instead of exact fit
fair-close limit: For the CFI, above this limit the model fit becomes close (good). For the RMSEA; below this limit the model fit becomes close (good)
R-Squared (optional)
Explained variance in dependent variables by their predictors
Average variance extracted (optional)
Amount of variance captured by a construct (latent variable)
Heterotrait-Monotrait Ratio (optional)
Correlations of constructs, that is, latent variables
Reliability (optional)
Coefficient omega: The closer to 1 the better. The total value denotes omega_t.
Mardias coefficients:
X% Confidence interval: lower and upper bound
Regression coefficients: Outcome and predictor
Standardized residuals covariance matrix
Modification indices (optional):
sepc: expected parameter change for standardized estimates: lv, all, and no exogenous covariates (nox)
Path diagram (optional)
Sensitivtiy model:
Sensitivity parameters that led to a change in significance
Sensitivity parameters: The path estimates for the phantom variable(s)
Summary of sensitivity parameters
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.