A Unified Approach to Two-Level Structural Equation Models and Linear Mixed Effects Models

  • Peter M. Bentler
  • Jiajuan Liang
Part of the Lecture Notes in Statistics book series (LNS, volume 192)


Two-level structural equation models (two-level SEM for simplicity) are widely used to analyze correlated clustered data (or two-level data) such as data collected from students (level-1 units) nested in different schools (level-2 units), or data collected from siblings (level-1 units) nested in different families (level-2 units). These data are usually collected by two sampling steps: randomly choosing some level-2 units; and then, randomly choosing some level-1 units from each chosen level-2 unit. Data collected in this way can be considered to be affected by two different random sources or random effects, namely, level-1 effects and level-2 effects. The substantive goal with such two-level data is to obtain theoretically meaningful and statistically adequate submodels for both the level-1 and level-2 effects. Realization of this main task consists of three steps: (1) set up an initial model with both level-1 and level-2 effects; (2) estimate the unknown model parameters; and (3) test the goodness-of-fit of the given model.


Linear Mixed Effect Model Saturated Model Linear Mixed Effect Model Bayesian Model Selection Unknown Model Parameter 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.



This work was supported by National Institute on Drug Abuse Grants DA01070, DA00017, and the University of New Haven 2006 and 2007 Summer Faculty Fellowships.


  1. Akaike, H. (1987). Factor analysis and AIC. Psychometrika, 52, 317–332MathSciNetMATHCrossRefGoogle Scholar
  2. Bauer, D. J. (2003). Estimating multilevel linear models as structural models. Journal of Educational and Behavioral Statistics, 28, 135–167CrossRefGoogle Scholar
  3. Bentler, P. M. (2006). EQS 6 Structural Equations Program Manual. Encino, CA: Multivariate Software ( Scholar
  4. Bentler, P. M. and Liang, J. (2003). Two-level mean and covariance structures: maximum likelihood via an EM algorithm. In S. Reise and N. Duan (Eds.), Multilevel Modeling: Methodological Advances, Issues, and Applications (pp. 53–70). Mahwah, NJ: Lawrence Erlbaum AssociatesGoogle Scholar
  5. Bentler, P. M., Liang, J. and Yuan, K.-H. (2005). Some recent advances in two-level structural equation models: estimation, testing and robustness. In J. Fan and G. Li (Eds.), Contemporary Multivariate Analysis and Experimental Design (pp. 211–232). NJ: World Scientific PublisherGoogle Scholar
  6. Bozdogan, H. (1987). Model selection and Akaike's information criterion (AIC): the central theory and its analytical extensions. Psychometrika, 52, 345–370MathSciNetMATHCrossRefGoogle Scholar
  7. Browne, M. W. and Cudeck, R. (1989). Single sample cross-validation indices for covariance structures. Multivariate Behavioral Research, 24, 445–455CrossRefGoogle Scholar
  8. Burnham, K. P. and Anderson, D. R. (2004). Multimodal inference: understanding AIC and BIC in model selection. Sociological Methods & Research, 33, 261–304MathSciNetCrossRefGoogle Scholar
  9. Cudeck, R. and Browne, M. W. (1983). Cross-validation of covariance structures. Multivariate Behavioral Research, 18, 147–157CrossRefGoogle Scholar
  10. Curran, P. J. (2003). Have multilevel models been structural equation models all along? Multivari-ate Behavioral Research, 38, 529–569CrossRefGoogle Scholar
  11. Dempster, A. P., Laird, N. M. and Rubin, D. B. (1977). Maximum likelihood from incomplete data via EM algorithm (with discussion). Journal of the Royal Statistical Society (Series B), 39, 1–38MathSciNetMATHGoogle Scholar
  12. du Toit, M. and du Toit, S. (2001). Interactive LISREL: User's Guide. Chicago: Scientific Software International, IncGoogle Scholar
  13. Gelman, A. and Rubin, D. B. (1995). Avoiding model selection in Bayesian social research. In A. E. Raftery (Ed.), Sociological Methodology (pp. 165–173). Oxford, UK: BlackwellGoogle Scholar
  14. Hoadley, B. (1971). Asymptotic properties of maximum likelihood estimators for the independent not identically distributed case. The Annals of Mathematical Statistics, 42, 1977–1991MathSciNetMATHCrossRefGoogle Scholar
  15. Kuha, J. (2004). AIC and BIC: comparison of assumptions and performance. Sociological Methods & Research, 33, 188–229MathSciNetCrossRefGoogle Scholar
  16. Lange, K. (1995a). A gradient algorithm locally equivalent to the EM algorithm. Journal of the Royal Statistical Society (Series B), 57, 425–437MATHGoogle Scholar
  17. Lange, K. (1995b). A Quasi-Newton acceleration of the EM algorithm. Statistica Sinica, 5, 1–18MathSciNetMATHGoogle Scholar
  18. Lee, S. Y. (1990). Multilevel analysis of structural equation models. Biometrika, 77, 763–772MathSciNetMATHCrossRefGoogle Scholar
  19. Lee, S. Y. and Poon, W. Y. (1998). Analysis of two-level structural equation models via EM type algorithms. Statistica Sinica, 8, 749–766MathSciNetMATHGoogle Scholar
  20. Lee, S.-Y. and Song, X.-Y. (2003). Bayesian model selection for mixtures of structural equation models with an unknown number of components. British Journal of Mathematical and Statistical Psychology, 56, 145–165MathSciNetCrossRefGoogle Scholar
  21. Liang, J. and Bentler, P. M. (2004). An EM algorithm for fitting two-level structural equation models. Psychometrika, 69, 101–122MathSciNetCrossRefGoogle Scholar
  22. McDonald, R. P. and Goldstein, H. (1989). Balanced versus unbalanced designs for linear structural relations in two-level data. British Journal of Mathematical and Statistical Psychology, 42, 215–232MathSciNetMATHCrossRefGoogle Scholar
  23. McLachlan, G. J. and Krishnan, T. (1997). The EM Algorithm and Extensions. New York: WileyMATHGoogle Scholar
  24. Mehta, P. D. and Neale, M. C. (2005). People are variables too: multilevel structural equations modeling. Psychological Methods, 10, 259–284CrossRefGoogle Scholar
  25. Meng, X.-L. (1998). Fast EM-type implementations for mixed effects models. Journal of the Royal Statistical Society (Series B), 60, 559–578MATHCrossRefGoogle Scholar
  26. Muthén, B. O. (1989). Latent variable modeling in heterogeneous populations. Psychometrika, 54, 557–585MathSciNetCrossRefGoogle Scholar
  27. Muthén, B. (1997). Latent variable modeling with longitudinal and multilevel data. In A. Raftery (Ed.), Sociological methodology (pp. 453–480). Boston: Blackwell PublishersGoogle Scholar
  28. Muthén, L. K. and Muthén, B. O. (2004). Mplus User's Guide. Los Angeles: Muthén & MuthénGoogle Scholar
  29. Raftery, A. E. (1995a). Bayesian model selection in social research. In A. E. Raftery (Ed.), Sociological Methodology (pp. 111–164). Oxford, UK: BlackwellGoogle Scholar
  30. Raftery, A. E. (1995b). Rejoinder: model selection is unavoidable in social research. In A. E. Raftery (Ed.), Sociological Methodology (pp. 185–195). Oxford, UK: BlackwellGoogle Scholar
  31. Raudenbush, S. W. (1995). Maximum likelihood estimation for unbalanced multilevel covariance structure models via the EM algorithm. British Journal of Mathematical and Statistical Psychology, 48, 359–370MATHCrossRefGoogle Scholar
  32. Rovine, M. J. and Molenaar, P. C. (2000). A structural modeling approach to a multilevel random coefficients model. Multivariate Behavioral Research, 35, 51–88CrossRefGoogle Scholar
  33. Skrondal, A. and Rabe-Hesketh, S. (2004). Generalized Latent Variable Modeling: Multilevel, Longitudinal and Structural Equation Models. Boca Raton, FL: Chapman and HillMATHCrossRefGoogle Scholar
  34. Song, X.-Y. and Lee, S.-Y. (2002). A Bayesian model selection method with applications. Computational Statistics & Data Analysis, 40, 539–557MathSciNetMATHCrossRefGoogle Scholar
  35. Sörbom, D. (1989). Model modification. Psychometrika, 54, 371–384MathSciNetCrossRefGoogle Scholar
  36. Yuan, K.-H. (2005). Fit indices versus test statistics. Multivariate Behavioral Research, 40, 115–148CrossRefGoogle Scholar
  37. Yuan, K.-H. and Bentler, P. M. (2003). Eight test statistics for multilevel structural equation models. Computational Statistics & Data Analysis, 44, 89–107MathSciNetMATHCrossRefGoogle Scholar
  38. Yuan, K.-H. and Bentler, P. M. (2006). Asymptotic robustness of standard errors in multilevel structural equation models. Journal of Multivariate Analysis, 97, 1121–1141MathSciNetMATHCrossRefGoogle Scholar
  39. Yuan, K.-H. and Bentler, P. M. (2007a). Structural equation modeling. In C. R. Rao and S. Sinharay (Eds.), Handbook of Statistics 26: Psychometrics (pp. 297–358). Amsterdam: North-HollandGoogle Scholar
  40. Yuan, K.-H. and Bentler, P. M. (2007b). Robust procedures in structural equation modeling. In S.-Y. Lee (Ed.), Handbook of Latent Variable and Related Models (pp. 367–397). Amsterdam: ElsevierCrossRefGoogle Scholar
  41. Yuan, K.-H., Marshall, L. L. and Weston, R. (2002). Cross-validation by downweighting influential cases in structural equation modeling. British Journal of Mathematical and Statistical Psychology, 55, 125–143CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC 2008

Authors and Affiliations

  1. 1.Los Angeles Departments of Psychology and StatisticsUniversity of CaliforniaLos Angeles
  2. 2.University of New Haven, College of BusinessWest Haven

Personalised recommendations