Multilevel latent covariate model for unbiased estimation of class-level climate–outcome relations

Research Method · All research methods
Evidence · 7 claims (7 for) · 3 studies (2 associational, 1 theoretical), q2–q3 · 2 of 3 report an effect size · 7 claims rest on one study

Description

The article recaps Lüdtke et al.'s (2008) multilevel latent covariate model, in which students' ratings of classroom climate are latently aggregated to a latent class mean that is regressed on an outcome at the class level. The model is presented as "the method of choice because it allows researchers to obtain unbiased estimates of the relations between classroom climate and outcome variables even when the ICC(2) values are not very high." It applies to reflective group-level constructs such as student-rated classroom management or teacher support.

Accounts

• The multilevel latent covariate model as the method of choice for unbiased estimation of class-level climate-outcome relations from student ratings: The article recaps Lüdtke et al.'s (2008) multilevel latent covariate model, in which students' ratings of classroom climate are latently aggregated to a latent class mean that is regressed on an outcome at the class level. The model is presented as "the method of choice because it allows researchers to obtain unbiased estimates of the relations between classroom climate and outcome variables even when the ICC(2) values are not very high." It applies to reflective group-level constructs such as student-rated classroom management or teacher support. (Steffen Zitzmann et al. (2021))
• Doubly latent models for compositional effects: Doubly latent models are multilevel latent variable models that estimate compositional effects while correcting for two sources of error: measurement error in individual-level measures, controlled "by using multiple indicators," and sampling error in aggregates, "adjusted for by using latent rather than a manifest aggregation to form the group-level aggregates." The article applies multi-group doubly latent models in Mplus 8 to SIMS80 data to test peer spillover and BFLPE effects, showing they change substantive conclusions relative to manifest models. (Televantou et al. (2023))
• Doubly latent multilevel structural equation modeling for climate-effect research: The article employs doubly latent multilevel SEM (Marsh et al., 2009) to test relationships between teacher and classroom constructs at the correct level of analysis. Its stated advantages are "the ability to control for measurement error at both L1 and L2 levels by using multiple indicators of a construct, as well as the ability to control for sampling error through the aggregating individual student responses to represent teacher level characteristic". It treats instructional quality as a classroom climate effect with the class, not the individual student, as referent. (Burić et al. (2020))

Claims

• The variability of the between slope estimate in the multilevel latent covariate model decreases with more classes and students, with higher ICC(1) of the ratings and larger slopes, but increases with higher ICC(1) of the outcome [+W]
• Under a fixed budget, a higher ICC(1) of the outcome variable shifts the optimal design toward more classes and fewer students per class and reduces maximum power [+W]
• A higher ICC(1) of students' ratings and a larger standardized between slope increase power, with opposite effects on the optimal split between classes and students per class [+W]
• Positive class-level peer spillover effects on mathematics achievement fail to replicate once measurement error is controlled with doubly latent models [+M]
• The Big-Fish-Little-Pond effect on mathematics self-concept is negative and statistically significant in all three countries under a fully specified doubly latent model [+M]
• Class-level instructional quality (classroom management, cognitive activation, supportive climate) is positively related to student self-efficacy [+M]
• Teacher self-efficacy is not directly related to student self-efficacy or intrinsic motivation when teacher and student reports are combined at the class level [+M]

Related Research Methods

• Multilevel latent variable model (ML-LVM) of judgment accuracy
• Multilevel conceptualization of autonomy support: climate, contextual, and perceived-difference constructs

Key Sources

• Steffen Zitzmann, Wolfgang Wagner, Martin Hecht, Christoph Helm, Christian Fischer, Lisa Bardach, Richard Göllner. (2021). How Many Classes and Students Should Ideally be Sampled When Assessing the Role of Classroom Climate via Student Ratings on a Limited Budget? An Optimal Design Perspective. Educational Psychology Review (2022) 34:511–536. https://doi.org/10.1007/s10648-021-09635-4
• Televantou, I., Marsh, H. W., Xu, K. M., Guo, J., & Dicke, T. (2023). Peer Spillover and Big-Fish-Little-Pond Effects with SIMS80: Revisiting a Historical Database Through the Lens of a Modern Methodological Perspective. Educational Psychology Review. https://doi.org/10.1007/s10648-023-09816-3
• Burić, Irena and Kim, Lisa (2020) Teacher self-efficacy, instructional quality, and student motivational beliefs: An analysis using multilevel structural equation modeling. Learning and Instruction. https://doi.org/10.1016/j.learninstruc.2019.101302

Page metadata

Field	Value
type	research-method
id	multilevel-latent-covariate-model-for-unbiased-estimation-of-class-level-climateoutcome-re
title	Multilevel latent covariate model for unbiased estimation of class-level climate–outcome relations
description	The article recaps Lüdtke et al.'s (2008) multilevel latent covariate model, in which students' ratings of classroom climate are latently aggregated to a latent class mean that is regressed on an outcome at the class level.
status	review
generated	by process:settle-candidates on 2026-10-21
trust	unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.