Variable names and data column names must not contain periods. Rename them consistently, for example x.1 to x_1, and check that the new names are unique. Study, group, and cluster values may contain periods.
It conducts the first stage analysis of TSSEM by pooling
correlation/covariance matrices. tssem1FEM() and tssem1REM()
use fixed- and random-effects models, respectively. tssem1() is a
wrapper of these functions.
The fixed-effects summary reports RMSEA using total sample size N as the
denominator; see summary for the formula.
tssem1(Cov, n, method = c("REM", "FEM"), cor.analysis = TRUE,
cluster = NULL, RE.type = c("Diag", "Symm", "Zero",
"User"), RE.startvalues = 0.1, RE.lbound = 1e-10,
RE.constraints = NULL, I2 = "I2q", acov = c("weighted",
"individual", "unweighted"), asyCovOld = FALSE,
model.name = NULL, suppressWarnings = TRUE, silent = TRUE,
run = TRUE, ...)
tssem1FEM(Cov, n, cor.analysis = TRUE, model.name = NULL,
cluster = NULL, suppressWarnings = TRUE, silent = TRUE,
run = TRUE, ...)
tssem1REM(Cov, n, cor.analysis = TRUE, RE.type = c("Diag",
"Symm", "Zero", "User"), RE.startvalues = 0.1,
RE.lbound = 1e-10, RE.constraints = NULL, I2 = "I2q",
acov = c("weighted", "individual", "unweighted"),
asyCovOld = FALSE, model.name = NULL, suppressWarnings = TRUE,
silent = TRUE, run = TRUE, ...)Either an object of class tssem1FEM for fixed-effects TSSEM,
an object of class tssem1FEM.cluster for fixed-effects TSSEM with
cluster argument, or an object of class tssem1REM for
random-effects TSSEM.
A list of correlation/covariance matrices. For both TSSEM methods, named matrices must all have unique row and column names for the same variables; each study is aligned to the first study's variable order before pooling and calculating sampling covariances. If all matrices are unnamed, their positions are used. Fixed-effects TSSEM requires complete correlations among the variables present in each study; random-effects TSSEM can handle partially missing correlations. It rejects a reported statistic whose sampling variance is unavailable rather than silently removing that statistic. Supply the missing information or choose an appropriate pooled working matrix. For correlation analysis, reported covariances require both corresponding variances so they can be standardized. Fixed-effects covariance analysis retains studies reporting only a variance. Fixed-effects correlation analysis rejects studies with no observed correlations; covariance analysis rejects studies with no observed variances.
A vector of sample sizes
Either "REM" (default if missing) or "FEM". If
it is "REM", a random-effects meta-analysis will be applied. If it is "FEM",
a fixed-effects meta-analysis will be applied.
Logical. The output is either a pooled correlation or a covariance matrix.
A character vector in tssem3L1 and tssemRobust1
or a vector of characters or numbers indicating the clusters in
tssem1. Analyses will be conducted for each cluster. It will be
ignored when method="REM".
Either "Diag", "Symm", "Zero" or
"User". If it is "Diag" (default if missing), a diagonal
matrix is used for the random effects meaning that the random effects are
independent. If it is "Symm", a symmetric matrix is used for the
random effects on the covariances among the correlation (or covariance)
vectors. If it is "Zero", there is no random effects which is similar
to the conventional Generalized Least Squares (GLS) approach to
fixed-effects analysis. If it is "User", the user has to specify the variance
component via the RE.constraints argument. This argument will be
ignored when method="FEM".
Starting values on the diagonals of the variance
component of the random effects. It will be ignored when
method="FEM".
Lower bounds on the diagonals of the variance component of
the random effects. It will be ignored when method="FEM".
A \(p*\) x \(p*\) matrix specifying the
variance components of the random effects, where \(p*\) is the number
of effect sizes. If the input is not a matrix, it is converted into a matrix
by as.matrix(). The default is that all covariance/variance
components are free. The format of this matrix follows
as.mxMatrix. Elements of the variance components can
be constrained equally by using the same labels. If a zero matrix is
specified, it becomes a fixed-effects meta-analysis.
Possible options are "I2q", "I2hm" and
"I2am". They represent the I2 calculated by using a typical
within-study sampling variance from the Q statistic, the harmonic mean and
the arithmetic mean of the within-study sampling variances (Xiong, Miller, &
Morris, 2010). More than one options are possible. If
intervals.type="LB", 95% confidence intervals on the heterogeneity
indices will be constructed.
If it is individual, the sampling variance-covariance
matrices are calculated based on individual correlation/covariance matrix.
If it is either unweighted or weighted (the default), the
average correlation/covariance matrix is calculated based on the unweighted
or weighted mean with the sample sizes. The average correlation/covariance
matrix is used to calculate the sampling variance-covariance matrices. This
argument is ignored with the method="FEM" argument.
Whether to use the legacy Hessian-based method. It requires complete covariances among retained variables in its working matrix. asyCov() is preferred. See asyCov.
A string for the model name in
mxModel.
Logical. If TRUE, warnings are suppressed. It
is passed to mxRun.
Logical. An argument to be passed to
mxRun
Logical. If FALSE, only return the mx model without
running the analysis.
Further arguments to be passed to mxRun
Mike W.-L. Cheung <[email protected]>
Cheung, M. W.-L. (2014). Fixed- and random-effects meta-analytic structural equation modeling: Examples and analyses in R. Behavior Research Methods, 46, 29-40.
Cheung, M. W.-L. (2013). Multivariate meta-analysis as structural equation models. Structural Equation Modeling, 20, 429-454.
Cheung, M. W.-L., & Chan, W. (2005). Meta-analytic structural equation modeling: A two-stage approach. Psychological Methods, 10, 40-64.
Cheung, M. W.-L., & Chan, W. (2009). A two-stage approach to synthesizing covariance matrices in meta-analytic structural equation modeling. Structural Equation Modeling, 16, 28-53.
wls, Cheung09,
Becker92, Digman97,
issp89, issp05