
A Function That Writes, Saves, and Exports Syntax for Fitting Correlated Dyadic Factor Models
Source:R/scriptCor.R
scriptCor.Rd
This function takes the outputted object from scrapeVarCross()
and automatically writes, returns, and exports (.txt
) lavaan
syntax
for specifying dyadic configural, loading, intercept, and residual invariant
two-factor models.
Arguments
- dvn
Input dvn list from
scrapeVarCross()
- scaleset
Input character to specify how to set the scale of the latent variable. Default is
"FF"
(fixed-factor; see Details for rationale), but user can specify"MV"
(Marker Variable)- lvname
Input character to (arbitrarily) name the latent variable in
lavaan
syntax- constr_dy_meas
Input character vector detailing which measurement model parameters to constrain across dyad members.
- constr_dy_struct
Input character vector detailing which structural model parameters to constrain across dyad members. Default is
c("variances", "means")
(in combination with defaults forconstr_dy_meas
, an indistinguishable correlated dyadic factors model), but user can specify any combination of"variances"
and"means"
, or"none"
.- writeTo
A character string specifying a directory path to where a
.txt
file of the resultinglavaan
script should be written. If set to“.”
, the.txt
file will be written to the current working directory. The default isNULL
, and examples use a temporary directory created bytempdir()
.- fileName
A character string specifying a desired base name for the
.txt
output file. The default isNULL
. The specified name will be automatically appended with the.txt
file extension. If a file with the same name already exists in the user's chosen directory, it will be overwritten.
Details
By default, many
dySEM::
functions (includingscriptCor()
) default to a fixed-factor method of scale-setting, whereby the latent variance of a given factor is constrained to 1 for both partners in the configurally invariant model, and then one of these variances is freely estimated in subsequent models of the invariance testing sequence. We have selected this default for two reasons: (1) the selection of a marker-variable is usually arbitrary, yet can have a large influence on the estimation and testing of of structural parameters (see https://stats.stackexchange.com/questions/402133/in-cfa-does-it-matter-which-factor-loading-is-set-to-1/402732#402732); and (2) the selection of a non-invariant marker-variable can have disastrous down-stream consequences for the identification of non-invariant measurement parameters, following a the rejection of an omnibus invariance constraint set (see Lee, Preacher, & Little, 2011).
See also
scrapeVarCross
which this function relies on.
Other uni-construct script-writing functions:
scriptUni()
Examples
dvn <- scrapeVarCross(
dat = commitmentQ,
x_order = "spi",
x_stem = "sat.g",
x_delim1 = ".",
x_delim2="_",
distinguish_1="1",
distinguish_2="2"
)
sat.indist.script <- scriptCor(
dvn,
scaleset = "FF",
lvname = "Sat"
)
sat.lvars.script <- scriptCor(
dvn,
scaleset = "FF",
lvname = "Sat",
constr_dy_meas = "loadings",
constr_dy_struct = "variances"
)
sat.resids.script <- scriptCor(
dvn,
scaleset = "FF",
lvname = "Sat",
constr_dy_meas = c("loadings", "intercepts", "residuals"),
constr_dy_struct = "none",
writeTo = tempdir(),
fileName = "dCor_residual"
)
sat.ints.script <- scriptCor(
dvn,
scaleset = "FF",
lvname = "Sat",
constr_dy_meas = c("loadings", "intercepts"),
constr_dy_struct = "none",
writeTo = tempdir(),
fileName = "dCor_intercept"
)
sat.loads.script <- scriptCor(
dvn,
scaleset = "FF",
lvname = "Sat",
constr_dy_meas = c("loadings"),
constr_dy_struct = "none",
writeTo = tempdir(),
fileName = "dCor_loading"
)
sat.config.script <- scriptCor(
dvn,
scaleset = "FF",
lvname = "Sat",
constr_dy_meas = "none",
constr_dy_struct = "none",
writeTo = tempdir(),
fileName = "dCor_configural"
)