cbsem (version 1.0.0)

gscals: Estimating GSC models belonging to scenarios reflective-reflective, formative-reflective and formative-formative

Description

gscals estimates GSC models alternating least squares. This leads to estimations of weights for the composites and an overall fit measure.

Usage

gscals(dat, B, indicatorx, indicatory, loadingx = FALSE, loadingy = FALSE,
  maxiter = 200, biascor = FALSE)

Arguments

dat

(n,p)-matrix, the values of the manifest variables. The columns must be arranged in that way that the components of refl are (absolutely) increasing.

B

(q,q) lower triangular matrix describing the interrelations of the latent variables: b_ij = 1 regression coefficient of eta_j in the regression relation in which eta_i is the depend variable b_ij = 0 if eta_i does not depend on eta_j in a direct way (b_ii = 0 !)

indicatorx

vector describing with which exogenous composite the X-variables are connected

indicatory

vector describing with which endogenous composite the Y-variables are connected

loadingx

logical TRUE when there are loadings for the X-variables in the model

loadingy

logical TRUE when there are loadings for the Y-variables in the model

maxiter

Scalar, maximal number of iterations

biascor

Boolean, FALSE if no bias correction is done, TRUE if parametric bootstrap bias correction is done.

Value

out list with components

Bhat	(q,q) lower triangular matrix with the estimated coefficients of the structural model
What	(n,q) matrix of weights for constructing the composites
lambdahat	vector of length p with the loadings or 0
iter	number of iterations used
fehl	maximal difference of parameter estimates for the last and second last iteration
composit	the data matrix of the composites
resid	the data matrix of the residuals of the structural model
S	the covariance matrix of the manifest variables
ziel	sum of squared residuals for the final sum
fit	The value of the fit criterion

Examples

# NOT RUN {
data(mobi250)
ind <- c(1, 1, 1, 4, 4, 4, 2, 2, 2, 3, 3, 5, 5, 5, 6, 6, 6, 7, 1, 1, 4, 4, 4, 4) 
o <- order(ind)
indicatorx <- c(1,1,1,1,1)
indicatory <- c(1, 1, 1, 2, 2, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 5, 5, 5)   
dat <- mobi250[,o]
dat <- dat[,-ncol(dat)]
B <- matrix(c(0,0,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,
              0,1,1,0,0,0,0,1,1,1,0,0,1,0,0,0,1,0),6,6,byrow=TRUE)
out <- gscals(dat,B,indicatorx,indicatory,loadingx=TRUE,loadingy=TRUE,maxiter=200,biascor=FALSE)

# }