Functions
Terrence edited this page May 10, 2022
·
34 revisions
-
clipboard()
: Copy the output of methods (e.g.,summary()
) forlavaan
objects into the clipboard, which can be pasted in other programs such as Excel -
saveFile()
: Save the attributes of thelavaan
object into a file -
compareFit()
: Compare fit measures across multiple nested or nonnestedlavaan
outputs
-
skew()
andkurtosis()
: Univariate skewness and excessive kurtosis -
mardiaSkew()
andmardiaKurtosis()
: Mardia's multivariate skewness and kurtosis -
mvrnonnorm()
: Convenience function to generate nonnormal data using Vale and Maurelli (1983) method
-
chisqSmallN()
: chi-squared test statistic (or difference) adjusted for small sample size -
moreFitIndices()
andnullRMSEA()
: Calculate more fit indices fromlavaan
objects: Gamma Hat (GFI*), Adjusted Gamma Hat (AGFI*), Corrected Akaike Information Criterion (AICc), Stochastic Information Criterion (SIC), Corrected Bayesian Information Criterion (BIC*), Hannan-Quinn Information Criterion (HQC), and the RMSEA of the null model -
miPowerFit()
: Model evaluation method provided by Satorra, Saris, & van der Weld (2009) that uses modification indices and the power of modification indices -
singleParamTest()
: Test each constraint that defines the differences between nested models
-
measurementInvariance()
andmeasurementInvarianceCat()
: Multiple-group measurement invariance Test for continuous and categorical indicators, respectively -
partialInvariance()
andpartialInvarianceCat()
: A range of partial invariance tests across multiple groups for continuous and categorical indicators, respectively -
longInvariance()
: Longitudinal Measurement Invariance Test. This function currently can test only one factor in different timepoints -
permuteMeasEq()
: Find the null hypothesis distribution of fit indices in measurement invariance across groups using the permutation method
-
SSpower()
: Power analysis for each parameter using Satorra and Bentler's (1985) method -
findRMSEApower()
: Find the power of rejecting bad models or retaining good models using RMSEA given sample size (MacCallum, Browne, & Suguwara, 1996) -
plotRMSEApower()
: Plot the power of rejecting bad models or retaining good models using RMSEA given a range of sample size -
plotRMSEAdist()
: Visualize the sampling distribution of RMSEA -
findRMSEAsamplesize()
: Find the minimum sample size given the desired power using RMSEA
-
findRMSEApowernested()
: Find the power of rejecting a pair of different models or retaining a pair of models similar models using a pair of RMSEA values given sample size (MacCallum, Browne, & Cai, 2006) -
plotRMSEApowernested()
: Plot the power of rejecting a pair of different models or retaining a pair of models similar models using a pair of RMSEA values given a range of sample size -
findRMSEAsamplesizenested()
: Find the minimum sample size given the desired power using a pair of RMSEA
-
auxiliary()
: Automatically add auxiliary variables to a lavaan model using full information maximum likelihood -
runMI()
: Fit structural equation models to multiply imputed data. The current function can optionally multiply impute missing values in the data, or users can feed the imputed data directly, which is recommended for users to have more control over the imputation process. The model is then fitted to each imputed dataset usinglavaanList()
, and results are returned in alavaan.mi
object. Methods forlavaan.mi
objects are available to pool standard results (see a list on theclass?lavaan.mi
help page). -
lavTestScore.mi()
andmodindices.mi()
: Analogous tolavTestScore()
andmodindices()
in thelavaan
package, but for pooling results across multiple imputations (i.e., forlavaan.mi
objects) -
twostage()
: Two-stage maximum likelihood in SEM. The current function will implement 2-stage ML (optionally using auxiliary variables) using lavaan to fit the model(s) in each step -
fmi()
: Find fractions of missing information (FMI) for summary statistics (means and (co)variances of continuous variables, thresholds and polychoric/polyserial correlations of ordered variables) from a single incomplete data set (usingmissing = "FIML"
) or a list of imputed data sets -
bsBootMiss()
: Model-based (Bollen-Stine) bootstrap with incomplete data -
quark()
andcombinequark()
: Principal component method to reduce the number of auxiliary variables for use in FIML estimation
-
indProd()
andorthogonalize()
: Creating indicator products without centering, with mean centering (Marsh, Wen, & Hau, 2004), with double mean centering (Lin et al., 2010), or with residual centering (Little, Bovaird, & Widaman, 2006) by all possible combinations or match-paired methods (Marsh et al., 2004) -
probe2WayMC()
: Probing two-way latent interaction with mean or double-mean centering -
probe3WayMC()
: Probing three-way latent interaction with mean or double-mean centering -
probe2WayRC()
: Probing two-way latent interaction with residual centering -
probe3WayRC()
: Probing three-way latent interaction with residual centering -
plotProbe()
: Plot the simple intercepts and slopes of latent interaction
-
efaUnrotate()
: Fitting an unrotated EFA model as alavaan
model -
orthRotate()
,oblqRotate()
, andfunRotate()
: Orthogonal or oblique rotation of standardized loadings obtained from the unrotated solution fitted byefaUnrotate()
-
ekc()
: Identify the number of factors to extract based on the Empirical Kaiser Criterion (EKC)
-
compRelSEM()
: Reliability of unit-weighted composites (e.g., sum scores or scale means) per construct. Also available for multidimensional, multilevel, and higher-order constructs, measured by continuous or categorical indicators (when enabled bylavaan
). Former functionsreliability()
andreliabiltyL2()
are now deprecated. -
maximalRelia()
: Maximal reliability, which is the reliability of weighted summed scores such that the weights provide the maximum value of reliability (Li, 1997).
-
parcelAllocation()
,PAVranking()
, andpoolMAlloc()
: Parcel allocation variability investigation (Sterba, 2011; Sterba & MacCallum, 2010; Sterba & Rights, 2016)
-
plausibleValues()
: Draw a distribution of plausible values of factor scores, to be treated as multiple imputations. -
imposeStart()
: Use parameter estimates from a lavaan output as starting values of another analysis model -
monteCarloCI()
: Mediation analysis using the Monte Carlo method (Selig & Preacher, 2012) -
splitSample()
: Randomly split samples into two different halves: a training and a (cross-)validation sample -
loadingFromAlpha()
: Estimate standardized factor loadings given a coefficient alpha when factor loadings are equally constrained (tau-equivalence) -
tukeySEM()
: Calculate Tukey's WSD post-hoc test for multiple group means comparison -
kd()
: Generate data based on the Kaiser-Dickman (1962) algorithm -
net()
: Test whether models estimated with ML are equivalent or nested -
htmt()
: Investigate the discriminant validity using Heterotrait-Monotrait Ratio