pith. sign in

arxiv: 1907.04148 · v1 · pith:WQ7VTIQOnew · submitted 2019-07-04 · 📊 stat.ME · stat.AP

Multiple membership multilevel models

George Leckie This is my paper

Pith reviewed 2026-05-25 08:47 UTC · model grok-4.3

classification 📊 stat.ME stat.AP
keywords multilevel modelsmultiple membershipnon-hierarchical datacross-classified structuressocial data analysisvariance componentsstatistical modeling
0
0 comments X

The pith

Multiple membership multilevel models extend standard multilevel models to analyze data where lower-level units belong to multiple higher-level units.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper sets out that traditional multilevel models assume each lower unit nests inside exactly one higher unit, such as a student in one school. Real data often violate this because units can belong to several higher units at once, for example students attending multiple schools or patients consulting multiple doctors. Multiple membership models address this by letting each lower unit contribute to several higher units through weights that sum to one. Ignoring the multiple memberships can distort estimates of group effects and variance components. The article therefore explains the data structures involved and shows how to specify and fit the corresponding models.

Core claim

Multiple membership multilevel models are an extension of standard multilevel models for non-hierarchical data that have multiple membership structures. Traditional multilevel models involve hierarchical data structures whereby lower-level units such as students are nested within higher-level units such as schools and where these higher-level units may in turn be nested within further groupings or clusters such as school districts, regions, and countries. With hierarchical data structures, there is an exact nesting of each lower-level unit in one and only one higher-level unit. However, social reality is more complicated than this, and so social and behavioural data often do not follow pure,

What carries the argument

Multiple membership data structures, in which each lower-level unit belongs to more than one higher-level unit and is assigned membership weights that sum to one.

If this is right

  • Analysts can obtain unbiased estimates of higher-level variance components even when units participate in several clusters.
  • Group-level effects such as school or doctor influences can be separated when the same lower-level unit contributes to multiple groups.
  • Model fit statistics and predictions improve for data generated by overlapping memberships rather than forced single memberships.
  • Software implementations become usable for the common practical case of students changing schools or patients switching providers.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same weighting approach could be applied to ecological data where individuals move between habitats.
  • Longitudinal extensions might let membership weights change over time to capture evolving group affiliations.
  • The distinction between cross-classified and multiple-membership structures suggests separate diagnostic checks before model choice.

Load-bearing premise

Social and behavioural data often do not follow pure or strict hierarchies.

What would settle it

A large-scale comparison showing that standard hierarchical multilevel models produce identical substantive conclusions to multiple membership models on every dataset with apparent multiple memberships would undermine the need for the extension.

read the original abstract

Multiple membership multilevel models are an extension of standard multilevel models for non-hierarchical data that have multiple membership structures. Traditional multilevel models involve hierarchical data structures whereby lower-level units such as students are nested within higher-level units such as schools and where these higher-level units may in turn be nested within further groupings or clusters such as school districts, regions, and countries. With hierarchical data structures, there is an exact nesting of each lower-level unit in one and only one higher-level unit. For example, each student attends one school, each school is located within one school district, and so on. However, social reality is more complicated than this, and so social and behavioural data often do not follow pure or strict hierarchies. Two types of non-hierarchical data structures which often appear in practice are cross-classified and multiple membership structures. In this article, we describe multiple membership data structures and multiple membership models which can be used to analyse them.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 1 minor

Summary. The manuscript describes multiple membership multilevel models as extensions of standard multilevel models designed to handle non-hierarchical data structures (specifically multiple membership and cross-classified structures) that arise when lower-level units such as individuals are not strictly nested within a single higher-level unit, using social and behavioural data as the motivating context.

Significance. If the characterizations are accurate, the paper supplies a clear conceptual overview that could aid applied researchers encountering violations of strict nesting; the absence of any derivation, simulation study, or empirical demonstration limits its novelty to exposition rather than methodological advance.

minor comments (1)
  1. [Abstract] The abstract states that 'social and behavioural data often do not follow pure or strict hierarchies' without supporting citation or prevalence data; adding a reference to existing literature on the frequency of such structures would strengthen the motivation.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive review and recommendation to accept. The referee's summary accurately captures the manuscript's purpose as a conceptual overview of multiple membership data structures and models.

Circularity Check

0 steps flagged

No circularity: purely descriptive methodological overview with no derivation chain

full rationale

The paper is an expository description of multiple membership multilevel models as extensions of standard multilevel models for non-hierarchical data. It contains no equations, no predictions, no fitted parameters presented as results, and no first-principles derivations. The central statements are definitional (e.g., 'Multiple membership multilevel models are an extension...') and background observations about data structures, with no load-bearing self-citations or reductions of claims to their own inputs. The paper is self-contained as a tutorial without any circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract provides no details on specific parameters, axioms, or new entities introduced.

pith-pipeline@v0.9.0 · 5676 in / 966 out tokens · 35178 ms · 2026-05-25T08:47:47.424496+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

19 extracted references · 19 canonical work pages

  1. [1]

    Beretvas, S. N. (2010). Cross-classified and multiple membership models. In J. J. Hox & J. K. Roberts (Eds.), Handbook of advanced multilevel analysis. London: Psychology Press

    work page 2010

  2. [2]

    Browne, W. J. (2019). MCMC estimation in MLwiN, v3.04. Centre for Multilevel Modelling, University of Bristol

    work page 2019

  3. [3]

    J., Goldstein, H., & Rasbash, J

    Browne, W. J., Goldstein, H., & Rasbash, J. (2001). Multiple membership multiple classification (MMMC) models. Statistical Modelling, 1, 103–124

    work page 2001

  4. [4]

    and Cameron, B

    Charlton, C., Rasbash, J., Browne, W.J., Healy, M. and Cameron, B. (2019) MLwiN Version 3.04. Centre for Multilevel Modelling, University of Bristol. MULTIPLE MEMBERSHIP MULTILEVEL MODELS 9

    work page 2019

  5. [5]

    Chung, H., & Beretvas, S. N. (2012). The impact of ignoring multiple membership data structures in multilevel models. British Journal of Mathematical and Statistical Psychology, 65, 185–200

    work page 2012

  6. [6]

    Fielding, A., & Goldstein, H. (2006). Cross-classified and multiple membership structures in multilevel models: An introduction and review. Research report RR791. London: Department for Education and Skills. URL: http://www/bristol.ac.uk/cmm/team/cross-classified-review.pdf

    work page 2006

  7. [7]

    Goldstein, H. (2011). Multilevel statistical models (4th ed.). London: Wiley

    work page 2011

  8. [8]

    W., & Goldstein, H

    Hill, P. W., & Goldstein, H. (1998). Multilevel modelling of educational data with cross- classification and missing identification of units. Journal of Educational and Behavioral Statistics, 23, 117–128

    work page 1998

  9. [9]

    B., Browne, W

    Lawson, A. B., Browne, W. J., & Vidal Rodeiro, C. L. (2003). Disease mapping using WinBUGS and MLwiN. London: Wiley

    work page 2003

  10. [10]

    Leckie, G. (2009). The complexity of school and neighbourhood effects and movements of pupils on school differences in models of educational achievement. Journal of the Royal Statistical Society: Series A, 172, 537–554

    work page 2009

  11. [11]

    Leckie, G. (2013). Multiple membership multilevel models - concepts. LEMMA VLE Module, 13, 1–61. URL: http://www.bristol.ac.uk/cmm/learning/course.html

    work page 2013

  12. [12]

    Leckie, G., & Charlton, C. (2013). runmlwin: A Program to Run the MLwiN Multilevel Modelling Software from within Stata. Journal of Statistical Software, 52, 1-40

    work page 2013

  13. [13]

    Rasbash, J., & Browne, W. (2001). Modelling non-hierarchical structures. In A. H. Leyland & H. Goldstein (Eds.), Multilevel modelling of health statistics. Chichester: Wiley

    work page 2001

  14. [14]

    Rasbash, J., & Browne, W. J. (2008). Non-hierarchical multilevel models. In J. De Leeuw & E. Meijer (Eds.), Handbook of multilevel analysis. New York: Springer. MULTIPLE MEMBERSHIP MULTILEVEL MODELS 10

    work page 2008

  15. [15]

    J., & Goldstein, H

    Rasbash, J., Steele, F., Browne, W. J., & Goldstein, H. (2019). A user’s guide to MLwiN, v3.03. Centre for Multilevel Modelling, University of Bristol

    work page 2019

  16. [16]

    W., & Bryk, A

    Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage Publications

    work page 2002

  17. [17]

    Snijders, T. A. B., & Bosker, R. J. (2012). Multilevel analysis: An introduction to basic and advanced multilevel modeling (2nd ed.). London: Sage

    work page 2012

  18. [18]

    C., Snijders, T

    Timmermans, A. C., Snijders, T. A. B., & Bosker, R. J. (2012). In search of value added in the case of complex school effects. Educational And Psychological Measurement, 73, 210-228

    work page 2012

  19. [19]

    Zhang, Z., Parker, R., Charlton, C., Leckie, G., & Browne, W. J. (2016). R2MLwiN - A program to run the MLwiN multilevel modelling software from within R. Journal of Statistical Software, 72, 10, 1-43. DOI: 10.18637/jss.v072.i10

    work page doi:10.18637/jss.v072.i10 2016