About this Research Topic
Multilevel modeling (MLM) as well as structural equation modeling (SEM) are commonly used in social and behavioral sciences. The main advantage of MLM is that complex relationships among variables can be studied on different levels as well as across different levels (see Hox, 2010; Snijders & Bosker, 2011; Raudenbrush & Bryk, 2002).
In addition, MLM is extremely flexible and can be used for a great variety of different data structures (e.g., cross-sectional data, longitudinal data, multirater data etc.). However, MLM is usually based on the analysis of manifest variables. As in single-level regression analysis, the regression weights in multilevel regression models can be substantially biased due to measurement error influences. The main advantage of SEM is to directly model the measurement error and investigate complex relationships among latent variables (see e.g., Bollen, 1989). Furthermore, the fit of a given structural equation model can be evaluated by different fit criteria (e.g., chi-square fit statistics, RMSEA, CFI etc.). Multilevel structural equation modeling (ML-SEM) combines the advantages of multi-level modeling and structural equation modeling and enables researchers to scrutinize complex relationships between latent variables on different levels (Mehta & Neale, 2005, Muthén, 1994). Considering the numerous advantages of ML-SEM, it is not surprising that many researchers are interested in the application as well as in the recent methodological developments of this new modeling framework.
In this Research Topic we aim to summarize and illustrate cutting-edge methodological developments in the field of multilevel structural equation modeling. Among these current hot topics, we focus on the adequacy and robustness of fit statistics in linear and non-linear ML-SEMs, Bayesian estimation techniques in ML-SEMs, causal inferences in ML-SEMs, mediation and moderation analysis in ML-SEMs. In addition to that, empirical applications for complex multilevel data structures are provided. For example, it is shown how multilevel data with categorical observed response variables, with cross-classified multilevel data, as well as with multitrait-multirater data can be analyzed.
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.