Bayesian estimation of single and multilevel models with latent variable interactions
Citations
42 citations
18 citations
Cites background from "Bayesian estimation of single and m..."
...Since version 8.3, Bayesian approaches have been available even for multilevel moderation and nonlinear effects (Asparouhov & Muthén, 2020)....
[...]
17 citations
Cites methods from "Bayesian estimation of single and m..."
...We ran this analysis using Bayesian estimation to decrease computational complexity (Asparouhov and Muthén, 2020)....
[...]
...Specifically, there are advantages of an alternative approach to within-level moderation analysis using latent interaction terms (Asparouhov and Muthén, 2020; Preacher et al., 2016)....
[...]
8 citations
7 citations
References
3,304 citations
Additional excerpts
...…the mean of a product EðXj XkÞ ¼ CovðXj;XkÞ: (5) Assuming multivariate normality for four random variables Xi, Xj, Xk, Xl any third-order moment about the mean (μ) is zero (see, e.g., Anderson, 1984), EððXi μiÞðXj μjÞðXk μkÞÞ ¼ 0; (6) while the fourth-order moment about the mean is a function…...
[...]
1,122 citations
"Bayesian estimation of single and m..." refers methods in this paper
...For ease of interpretation, the regression Equation (1) can be re-written in the equivalent form η3 ¼ ðβ1 þ β3 η2Þ η1 þ β2 η2 þ ζ3; (2) where ðβ1 þ β3 η2Þ is a moderator function (Klein & Moosbrugger, 2000) so that the β1 strength of influence of η1 on η3 is moderated by β3 η2....
[...]
...Interactions between two latent variables, however, do not lead to closed form expressions for the likelihood and cannot be estimated with the maximum-likelihood method without numerical integration, see Klein and Moosbrugger (2000)....
[...]
...In the single-level context, the maximum-likelihood (ML) estimation is described in Klein and Moosbrugger (2000)....
[...]
...Interactions between two latent variables, however, do not lead to closed form expressions for the likelihood and cannot be estimated with the maximum-likelihood method without numerical integration, see Klein and Moosbrugger (2000). In Mplus, the dimension of numerical integration corresponds to the number of latent variables used in the interaction terms....
[...]
...The method is asymptotically equivalent to the LMS method discussed in Klein and Moosbrugger (2000)....
[...]
1,045 citations
891 citations
"Bayesian estimation of single and m..." refers methods in this paper
...For an overview of the ML approach and various estimators suggested in earlier work, see Marsh et al. (2004)....
[...]
607 citations
"Bayesian estimation of single and m..." refers background or result in this paper
...These findings are in line with the findings in Lüdtke et al. (2008) and Asparouhov and Muthén (2019) regarding latent variable centering for predictors in multilevel modeling....
[...]
...E(η3) = β1 0 + β2 0 + β3 E(η1 η2) + 0 (12) = β3 Cov(η1, η2)....
[...]
...That bias is generally referred to as Lüdtke’s bias, see Lüdtke et al. (2008) and Asparouhov and Muthén (2019)....
[...]