A general multilevel SEM framework for assessing multilevel mediation
The DOI matches its Crossref record; 149 of the 150 references it cites were found in a registry or an index.
Abstract
Several methods for testing mediation hypotheses with 2-level nested data have been proposed by researchers using a multilevel modeling (MLM) paradigm. However, these MLM approaches do not accommodate mediation pathways with Level-2 outcomes and may produce conflated estimates of between- and within-level components of indirect effects. Moreover, these methods have each appeared in isolation, so a unified framework that integrates the existing methods, as well as new multilevel mediation models, is lacking. Here we show that a multilevel structural equation modeling (MSEM) paradigm can overcome these 2 limitations of mediation analysis with MLM. We present an integrative 2-level MSEM mathematical framework that subsumes new and existing multilevel mediation approaches as special cases. We use several applied examples and accompanying software code to illustrate the flexibility of this framework and to show that different substantive conclusions can be drawn using MSEM versus MLM.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“However, half of these studies pair two or more employees with the same supervisor (i.e.nested data), and use traditional regression methods to analyze the data, which is inappropriate, as the assumption of independence of observations is violated (Preacher, Zyphur, & Zhang, 2010).”
From Servant Leadership: A systematic review and call for future research (2019) · cites this paper as Preacher, Zyphur, & Zhang, 2010
In the citing paper
“However, half of these studies pair two or more employees with the same supervisor (i.e.nested data), and use traditional regression methods to analyze the data, which is inappropriate, as the assumption of independence of observations is violated (Preacher, Zyphur, & Zhang, 2010).”
From Servant Leadership: A systematic review and call for future research (2019) · cites this paper as Preacher, Zyphur, & Zhang, 2010
In the citing paper
“Multi-level analyses take into consideration the lack of independence in ratings made by leaders and followers as well as address within versus between leader effects (Preacher et al., 2010).”
From Servant Leadership: A systematic review and call for future research (2019) · cites this paper as Preacher et al., 2010
Citations
Cites
Baron, R. M., & Kenny, D. A. (1986) · Journal of Personality and Social Psychology · doi:10.1037/0022-3514.51.6.1173 · Cited by 91,532 · cited 2×
“…(Wood, Goodman, Beckmann, & Cook, 2008), following the instructions of numerous methodologists who have investigated mediation analysis (e.g., Baron & Kenny, 1986; Edwards & Lambert, 2007; MacKinnon, 2008; MacKinnon et al., 2002; Preacher & Hayes, 2004, 2008a, 2008b; Shrout & Bolger, 2002).”
Self-efficacy: The exercise of control
Bandura, A., Freeman, W. H., & Lightsey, R. (1999) · Journal of Cognitive Psychotherapy · doi:10.1891/0889-8391.13.2.158 · Cited by 46,835
Preacher, K. J., & Hayes, A. F. (2008) · Behavior Research Methods · doi:10.3758/brm.40.3.879 · Cited by 34,117 · cited 2×
“…(Wood, Goodman, Beckmann, & Cook, 2008), following the instructions of numerous methodologists who have investigated mediation analysis (e.g., Baron & Kenny, 1986; Edwards & Lambert, 2007; MacKinnon, 2008; MacKinnon et al., 2002; Preacher & Hayes, 2004, 2008a, 2008b; Shrout & Bolger, 2002).”
Self-citationHierarchical Linear Models: Applications and Data Analysis Methods.
Longford, N. T., Bryk, A. S., & Raudenbush, S. W. (1993) · Contemporary Sociology · doi:10.2307/2075823 · Cited by 18,724 · cited 1×
“These recommended procedures for testing multilevel mediation have been developed and framed almost exclusively within the standard multilevel modeling (MLM) paradigm (for thorough treatments of MLM, see Raudenbush & Bryk, 2002, and Snijders & Bosker, 1999) and implemented with commercially available MLM software, suc…”
SPSS and SAS procedures for estimating indirect effects in simple mediation models
Preacher, K. J., & Hayes, A. F. (2004) · Behavior Research Methods, Instruments, & Computers · doi:10.3758/bf03206553 · Cited by 17,383 · cited 1×
“…(Wood, Goodman, Beckmann, & Cook, 2008), following the instructions of numerous methodologists who have investigated mediation analysis (e.g., Baron & Kenny, 1986; Edwards & Lambert, 2007; MacKinnon, 2008; MacKinnon et al., 2002; Preacher & Hayes, 2004, 2008a, 2008b; Shrout & Bolger, 2002).”
Self-citationMultilevel analysis : an introduction to basic and advanced multilevel modeling
Snijders, T. A. B., & Bosker, R. (1999) · Cited by 13,775
Mediation in experimental and nonexperimental studies: New procedures and recommendations.
Shrout, P. E., & Bolger, N. (2002) · Psychological Methods · doi:10.1037/1082-989x.7.4.422 · Cited by 11,335 · cited 2×
“…(Wood, Goodman, Beckmann, & Cook, 2008), following the instructions of numerous methodologists who have investigated mediation analysis (e.g., Baron & Kenny, 1986; Edwards & Lambert, 2007; MacKinnon, 2008; MacKinnon et al., 2002; Preacher & Hayes, 2004, 2008a, 2008b; Shrout & Bolger, 2002).”
A comparison of methods to test mediation and other intervening variable effects
MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. (2002) · Psychological Methods · doi:10.1037/1082-989x.7.1.83 · Cited by 9,406 · cited 2×
“…(Wood, Goodman, Beckmann, & Cook, 2008), following the instructions of numerous methodologists who have investigated mediation analysis (e.g., Baron & Kenny, 1986; Edwards & Lambert, 2007; MacKinnon, 2008; MacKinnon et al., 2002; Preacher & Hayes, 2004, 2008a, 2008b; Shrout & Bolger, 2002).”
Confidence Limits for the Indirect Effect: Distribution of the Product and Resampling Methods
MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004) · Multivariate Behavioral Research · doi:10.1207/s15327906mbr3901_4 · Cited by 7,806 · cited 2×
“MacKinnon, Fritz, Williams, and Lockwood (2007); MacKinnon, Lockwood, and Williams (2004); and Williams and MacKinnon (2008) described the distribution of the product method for obtaining confidence limits for indirect effects.”
(2013) · Handbooks of Sociology and Social Research · doi:10.1007/978-94-007-6772-0 · Cited by 7,486
Cited by
Items 1,001–1,033 of 1,033
Thiel, C. E., McClean, S., Harvey, J., & Prince, N. (2023) · Journal of Organizational Behavior · doi:10.1002/job.2748 · Cited by 41
Mehmood, Q., Hamstra, M. R. W., Nawab, S., & Vriend, T. (2016) · Journal of Occupational and Organizational Psychology · doi:10.1111/joop.12153 · Cited by 39
“We used Mplus 7.31 (Muth (cid:2) en & Muth (cid:2) en, Los Angeles, CA, USA) to test our hypotheses in a multilevel SEM path model (Preacher, Zyphur, & Zhang, 2010) in which we separated within-and between-level components of our (indirect) effects (Tables 1 and 2).”
Influential(2015) · doi:10.1097/01.j.pain.0000460324.18138.0a · Cited by 38
“1) was tested according to the recommendations provided by Preacher et al.(24) Consistent with the MSEM approach, all paths were specified to have random intercepts and fixed slopes with 1 exception; the relation between ratings of last night’s sleep quality and end-of-day activity”
MethodInfluentialZhang, Y., Liu, X., & Chen, W. (2020) · Journal of Business and Psychology · doi:10.1007/s10869-019-09650-x · Cited by 35
Sun, S., Burke, M., Chen, H., Tan, Y. (., Zhang, J., & Hou, L. (2022) · Human Relations · doi:10.1177/0018726721992849 · Cited by 34
“The 90% CI corresponds to a one-tailed, α = 0.05 hypothesis test, which is justified in mediation research and given a priori hypothesis (Preacher et al., 2010: 217).”
MethodInfluentialSchenke, K., Nguyen, T., Watts, T. W., Sarama, J., & Clements, D. H. (2017) · Journal of Educational Psychology · doi:10.1037/edu0000165 · Cited by 34
“This method has advantages over the Sobel test in that the distribution of the indirect effect is not assumed to be normally distributed (Preacher et al., 2010).”
Predictors and social consequences of daily pain expectancy among adults with chronic pain
Mun, C. J., Thummala, K., Davis, M. C., Karoly, P., Tennen, H., & Zautra, A. J. (2017) · Pain · doi:10.1097/j.pain.0000000000000903 · Cited by 34
“Third, MSEM can account for measurement error.(50) Multilevel structural equation modeling automatically partitions the within– and between– person-level variances; thus centering each study variable is not necessary prior to analyses and the coefficients can be directly interpreted at the corresponding levels of anal…”
BackgroundMethodInfluentialSharifian, N., & Zahodne, L. B. (2021) · The Journal of General Psychology · doi:10.1080/00221309.2020.1743228 · Cited by 34
“ICCs revealed sufficient variance at the within- and between-person levels for a multilevel model approach across Level-1 variables.”
MethodPruessner, L., Timm, C., Kalmar, J., Bents, H., Barnow, S., & Mander, J. (2024) · Depression and Anxiety · doi:10.1155/2024/9081139 · Cited by 37
Yang, C., Yu, R., Hu, X., Luo, L., Huang, T. S.-T., & Shanks, D. R. (2021) · Metacognition and Learning · doi:10.1007/s11409-020-09254-4 · Cited by 29
“Multilevel mediation is becoming increasingly popular in cognitive psychology, and more and more software packages for conducting such analyses are available to researchers (e.g., Muthén and Muthén 1998-2010; Preacher et al. 2010; Vuorre 2017; Zhang et al. 2008).”
MethodBeidas, R. S., Volpp, K. G., Buttenheim, A. N., Marcus, S. C., Olfson, M., Pellecchia, M., Stewart, R. E., Williams, N. J., Becker-Haimes, E. M., Candon, M., et al. (2019) · JMIR Research Protocols · doi:10.2196/12121 · Cited by 28
“This work will build on and extend 2 approaches to test indirect effects in multilevel models—multilevel structural equation modeling [73] and the centered within context with means reintroduced approach [65]—to address 3-level models with mediators, interventions, and randomization at different levels.”
MethodMaking nice or faking nice? Exploring supervisors’ two-faced response to their past abusive behavior
McClean, S. T., Courtright, S. H., Yim, J., & Smith, T. A. (2021) · Personnel Psychology · doi:10.1111/peps.12424 · Cited by 28
Taylor, S. S., Davis, M. C., Yeung, E. W., Zautra, A. J., & Tennen, H. A. (2017) · Journal of Behavioral Medicine · doi:10.1007/s10865-016-9811-1 · Cited by 27
“FIML via an accelerated equation modeling algorithm procedure used in Mplus version 7 (Muthén & Muthén, 1999–2012) is robust to missing data, non-normality, and unbalanced cluster sizes in data (Muthén & Asparouhov, 2008; Preacher et al., 2010).”
MethodWest, T. H. R., Daher, P., Dawson, J. F., Lyubovnikova, J., Buttigieg, S. C., & West, M. A. (2022) · BMJ Open · doi:10.1136/bmjopen-2021-052778 · Cited by 27
Vowles, K. E., Sowden, G., Hickman, J., & Ashworth, J. (2019) · Behaviour Research and Therapy · doi:10.1016/j.brat.2018.10.012 · Cited by 24
Salvy, S.-J., Feda, D. M., Epstein, L. H., & Roemmich, J. N. (2017) · Preventive Medicine Reports · doi:10.1016/j.pmedr.2017.04.009 · Cited by 24
“Consequently, multilevel models were used to estimate parameters in the presence of clustering, with random intercepts at the family level (Pituch et al., 2006; Preacher et al., 2010; Wang et al., 2011), using PROC MIXED models in the SAS software, Version 9.4 (SAS, 2013).”
MethodInfluentialO'Toole, M. S., Mennin, D. S., Hougaard, E., Zachariae, R., & Rosenberg, N. K. (2015) · Clinical Psychology & Psychotherapy · doi:10.1002/cpp.1926 · Cited by 24
Shaping employee green behavior: a multilevel approach with Pygmalion effect
Mo, Z., Liu, M. T., & Wu, P. (2022) · Asia Pacific Journal of Marketing and Logistics · doi:10.1108/apjml-07-2020-0473 · Cited by 24
Interrater reliability for multilevel data: A generalizability theory approach.
ten Hove, D., Jorgensen, T. D., & van der Ark, L. A. (2022) · Psychological Methods · doi:10.1037/met0000391 · Cited by 29
Daks, J. S., Peltz, J. S., & Rogge, R. D. (2022) · Journal of Contextual Behavioral Science · doi:10.1016/j.jcbs.2022.08.011 · Cited by 22
Networks of prospective thoughts: The organisational role of emotion and its impact on well-being
Demblon, J., & D'Argembeau, A. (2016) · Cognition and Emotion · doi:10.1080/02699931.2015.1015967 · Cited by 21
Devine, R. T., Grumley Traynor, I., Ronchi, L., & Lecce, S. (2024) · Child Development · doi:10.1111/cdev.14085 · Cited by 17
Christensen-Salem, A., Kinicki, A., Perrmann-Graham, J., & Walumbwa, F. O. (2023) · Human Relations · doi:10.1177/00187267221119767 · Cited by 16
“We tested the significance of the indirect effects using a parametric bootstrap procedure (Preacher et al., 2010).”
MethodSciffer, M. G., Perry, L. B., & McConney, A. (2022) · Large-scale Assessments in Education · doi:10.1186/s40536-022-00142-8 · Cited by 16
Hofheinz, C., Reder, M., & Michalak, J. (2020) · Psychotherapy Research · doi:10.1080/10503307.2019.1685138 · Cited by 15
“In a second step, multilevel SEM (MSEM) mediation analyses were conducted (Preacher, Zhang, & Zyphur, 2011; Preacher, Zyphur, & Zhang, 2010).”
MethodLi, Y. N., Cheng, B. H., Yu, B., & Zhu, J. N. Y. (2024) · Personnel Psychology · doi:10.1111/peps.12585 · Cited by 13
“At Level 1, we specified the effects of daily physical activity on the mediators with random slopes and other effects with fixed slopes (Preacher et al., 2010).”
Toniolo-Barrios, M., & ten Brummelhuis, L. L. (2023) · Personality and Individual Differences · doi:10.1016/j.paid.2023.112392 · Cited by 13
A Primer on Two-Level Dynamic Structural Equation Models for Intensive Longitudinal Data
McNeish, D. (2018) · doi:10.31234/osf.io/j56bm · Cited by 12
Not all forms of artificial intelligence are perceived equal: AI functions and work outcomes
Cao, J., Yao, J., Sun, S., Song, Z., & Zhang, F. (2025) · Journal of Open Innovation: Technology, Market, and Complexity · doi:10.1016/j.joitmc.2025.100521 · Cited by 10
Constantino, M. J., Romano, F. M., Coyne, A. E., Westra, H. A., & Antony, M. M. (2018) · Psychotherapy Research · doi:10.1080/10503307.2017.1301689 · Cited by 7
Mabbe, Soenens, Vansteenkiste, & Pauw, D. (2020) · Merrill-Palmer Quarterly · doi:10.13110/merrpalmquar1982.66.3.0308 · Cited by 5
“Data are suitable for multilevel structural equation modeling when the ICC is above .05 (Preacher et al., 2010).”
MethodMangelsdorf, S. N., Cagliarini, D., Lee, Y. Y., Mihalopoulos, C., Liu, V., Valentine, L., Bendall, S., Koval, P., D'Alfonso, S., Davey, C., et al. (2024) · BMJ Open · doi:10.1136/bmjopen-2024-088695 · Cited by 3
“51 Specifically, person-specific slopes representing change over time in the proposed mechanism- of-action variables will be examined as mediators of the effect of treatment group ( Rebound vs control) on risk of relapse.”
(2013) · Cited by 1
Metadata
- Authors
- Preacher, K. J., Zyphur, M. J., & Zhang, Z.
- Affiliations
- University of Kansas (US)
- The University of Melbourne (AU)
- Arizona State University (US)
- Venue
- Psychological Methods
- Published
- 2010
- Volume
- 15
- Issue
- 3
- Pages
- 209-233
- ISSN
- 1082-989X (ISSN-L)
- 1939-1463
- OpenAlex
- W2124596021
- Cited by
- 3,796
- References
- 150
- Citations by year
- 2026: 349 · 2025: 350 · 2024: 359 · 2023: 335 · 2022: 376 · 2021: 346 · 2020: 311 · 2019: 268 · 2018: 227 · 2017: 237 · 2016: 168 · 2015: 158 · 2014: 124 · 2013: 85 · 2012: 60
- FWCI
- 49.11
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Psychological Methods (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
- Social Sciences › Decision Sciences › Management Science and Operations Research › Psychometric Methodologies and Testing
- Keywords
- multilevel mediation, multilevel structural equation modeling, indirect effects, nested data
- MeSH terms
- Data Interpretation, Statistical, Humans, Psychology, Research Design, Models, Statistical, Causality, Behavioral Research, Multilevel Analysis
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.