Pith. sign in

REVIEW 3 major objections 5 minor 162 references

Statistical analysis of block structured latent variable models

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper proves that block structured latent variable models are structurally identifiable exactly when the M-Q Condition holds, and that the constrained maximum likelihood estimator then achieves oracle rates and efficient inference.

desk verdict A substantial theory paper with a real identifiability contribution, but the inference results rest on a global-optimality theorem that is deferred to an unavailable supplement. read the letter →

arxiv 2608.10328 v1 pith:BO4FHVPX submitted 2026-08-10 stat.ME math.STstat.TH

Chengyu Cui , Gongjun Xu This is my paper · ORCID
classification stat.MEmath.STstat.TH MSC 62H2562F1262E20
keywords blockstructuredlatentvariablemodelidentifiabilitymaximumlikelihoodestimationnon-convexoptimizationLagrangianformulationfactoranalysisoracleratesasymptoticnormality
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper studies latent variable models in which observed variables are partitioned into blocks, each block loading on a designated subset of latent factors, with optional orthogonality constraints between factors. It aims to replace case-by-case analyses of special designs with a single statistical theory: necessary and sufficient conditions for identifiability, consistency of the constrained maximum likelihood estimator, and efficient inference. The central claim is that identifiability holds exactly when the M-Q Condition is satisfied, meaning every single latent factor can be isolated by set intersections and differences of the block factor sets, aided by the orthogonality constraints. To make estimation tractable, the paper introduces a Lagrangian-type objective whose global minimizer coincides with the constrained maximum likelihood estimator, and a block-wise gradient algorithm that converges linearly to it. If the theory is correct, the practical question "which block design identifies my model?" becomes a checkable combinatorial test, and the resulting estimators inherit oracle error rates and Cramér-Rao optimal variances.

What carries the argument

The M-extended $\pi$-system $\pi_M(A_Q)$ is the smallest collection of factor-index subsets that contains each block's factor set $A_k$ and is closed under intersections and under a difference rule triggered by orthogonality entries of $M$; the M-Q Condition requires every singleton $\{r\}$ to belong to it. This closure is the combinatorial certificate of identifiability: it records exactly which contrasts among latent factors the block design and orthogonality constraints can distinguish. The second workhorse is the Lagrangian-type objective $L_\nu(F, \Lambda_Q, \beta) = L(F, \Lambda_Q, \beta) + \nu P(F)$, where $P(F)$ is a quadratic penalty vanishing exactly on the normalization and orthogonality constraints. Its scaled Hessian's smallest eigenvalue is shown to be governed by $J^*/J$, the relative size of the smallest block that cannot be removed while preserving the M-Q Condition, and this curvature bound drives the global-optimality, error-rate, and algorithm-convergence results.

What would settle it

Construct a design where two latent factors appear together in every block (both in or both out of each $A_k$) and no orthogonality constraint separates them; the M-Q Condition fails, so Theorem 1 predicts non-identifiability, and the likelihood should admit a continuum of equivalent parameterizations. The same design, simulated with $N$ and $J$ growing, should show the smallest eigenvalue of the scaled Hessian at the truth converging to zero, contradicting the strong-convexity conclusion.

Watch

Extended reading notes

Core claim

Theorem 1 states that the parameters $(F, \Lambda_Q, \beta)$ of a block structured latent variable model are structurally identifiable if and only if every singleton $\{r\}$ lies in the M-extended $\pi$-system generated by the block sets $A_Q$. Theorem 3 shows that, under the same condition together with regularity assumptions, the constrained maximum likelihood estimator equals the global minimizer of the Lagrangian-type objective $L_\nu$, and also equals the unique minimizer of $L_\nu$ in a small neighborhood of the true parameters. Theorems 4 and 5 give non-asymptotic $\ell_2$ and $\ell_\infty$ error bounds and asymptotic normality: loadings and intercepts estimate at rate $N^{-1} + J^{-2}$, latent factors at $N^{-2} + J^{-1}$, and the asymptotic variances reach the oracle Cramér-Rao lower bounds. The paper's claim, taken as a whole, is that identifiability, consistency, and efficiency for this model class are fully characterized by the combinatorial M-Q Condition and the size $J^*$ of the smallest non-removable block.

Load-bearing premise

The theory assumes every block is informative enough to identify all latent factors it loads on (full-rank loadings with eigenvalues bounded away from zero) and that the smallest non-removable block grows proportionally with the total number of items, so $J^*/J$ stays bounded away from zero.

Editorial extensions

If this is right

  • A practitioner can decide whether a proposed block design and orthogonality constraint identify the model by computing the closure $\pi_M(A_Q)$ and checking whether every singleton appears, instead of verifying analytic conditions design by design.
  • Given a fixed block structure, the minimal orthogonality constraint needed for identifiability can be found by relaxing entries of $M$ until the M-Q Condition fails.
  • The constrained maximum likelihood estimator can be computed by a first-order block-wise gradient algorithm with linear convergence, and after $T \gg \log(N \vee J)$ iterations the algorithm's output inherits the exact estimator's asymptotic distribution.
  • Across scaling regimes such as $N = o(J^2)$ and $J = o(N^2)$, loading and factor estimators reach oracle rates and asymptotically efficient variances, so uncertainty quantification for factor scores and loadings is available in broad settings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The closure criterion could be used prospectively as a design tool for multi-source data integration: arrange source groupings so each latent dimension is separable in $\pi_M(A_Q)$, avoiding identifiability failures before data collection.
  • The theory predicts a phase transition as $J^*/J \to 0$: curvature and error bounds degrade, so designs whose identifying block is vanishingly small should exhibit markedly worse estimation, which is testable in simulation.
  • The Lagrangian equivalence is a general template: any constrained non-convex estimator whose constraint set can be encoded as the zero set of a quadratic penalty may inherit the same global-optimum coincidence, provided a curvature certificate analogous to $J^*/J$ exists.
  • The paper treats the link function and the number of latent factors as known; extending the closure condition to unknown link or unknown dimension is a natural next step that the current framework does not yet cover.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This paper studies a general class of block structured latent variable models, where the observed variables are partitioned into blocks and each block loads on a specified subset of latent factors, with partial orthogonality constraints encoded by a matrix M. The main contributions are: (i) a combinatorial condition, called the M-Q Condition, claimed to be necessary and sufficient for structural identifiability of the model parameters (Theorem 1); (ii) a Lagrangian-type formulation with a quadratic penalty P(F), whose global optimum is claimed to coincide exactly with the constrained maximum likelihood estimator (Theorem 3); (iii) non-asymptotic l2 and l-infinity error bounds and asymptotic normality results with oracle Cramer-Rao variances (Theorems 4 and 5); and (iv) a blockwise first-order algorithm with linear convergence and statistical equivalence to the exact estimator (Proposition 1 and Corollary 1). The theory is illustrated by simulations over four block designs and by an application to PISA 2022 data with a bifactor model. All technical proofs are deferred to a separate Supplementary Material file that is not included in the arXiv submission.

Significance. If the results are correct, the paper would provide the first general, checkable characterization of identifiability for block structured latent variable models, and would unify many existing special-case results. The M-Q Condition is a genuinely new combinatorial device, and the claimed rates (N^{-1}+J^{-2} for loadings, N^{-2}+J^{-1} for factors) together with oracle-efficiency asymptotic distributions would be substantial advances over the existing O_p(N^{-1}+J^{-1}) bounds. The simulation coverage rates and the empirical analysis are consistent with the stated theorems and add credibility. However, the central equivalence in Theorem 3 is only sketched, and the entire proof machinery is in an unavailable supplement, so the paper as submitted cannot be fully verified.

major comments (3)
  1. [Section 3.2, Theorem 3 and Eq. (14)] The claimed exact equality between the Lagrangian minimizer and the constrained MLE is the lynchpin of the paper, but it is not established in the main text. Since P(F) is scaled by 1/(NJ) while the negative log-likelihood L has scale NJ, a finite-sample improvement of O(1) in L can dominate any penalty contribution; the proof must show that a zero-penalty representative exists with no increase in L. The text only states that a reference parameter set in Xi* has nearly minimal L, that the global minimizer therefore has small P, and that a constructive argument converts P != 0 into a strictly smaller L_nu. These are the exact steps that require Lemma 3, the Hv construction, and the P(~F) != 0 contradiction, all of which are deferred to the unavailable supplement. Because Theorems 4, 5, Proposition 1, and Corollary 1 all flow through Theorem 3, this is a load-bearing gap.
  2. [Assumption 3 and Theorem 2, Eq. (12)] The theory requires J* to be proportional to J (Assumption 3), and the strong convexity lower bound in Eq. (12) degrades with J*/J. But J* is defined as max over S of min_{k in S} |J_k|, which can be O(1) while the M-Q Condition still holds, because a single small block can supply the identifying restrictions. Thus the paper excludes designs whose identifying block is vanishingly small, and its claims in the introduction and Section 4 about covering a 'wide range of asymptotic regimes' and 'various block designs' are considerably stronger than what is actually proved. The statement should be explicitly qualified to the J*/J asymptotically non-vanishing regime.
  3. [Throughout; Supplement references] The arXiv version submitted for review does not include the Supplementary Material, yet the proofs of all main results are in it: Lemma 1 and Lemma 3 are used in Section 3, the proof details of Theorem 3 are in Section E, Proposition 1 states 'Proof see Section E.5', and the initialization of the algorithm is given as Algorithm S2 in the supplement. Consequently, none of the theorem statements can be independently checked from the submitted document. This is not itself a mathematical error, but it makes the manuscript incomplete as a standalone submission; the supplement must be provided to the referees.
minor comments (5)
  1. [Definition 2] The second closure rule of the M-extended pi-system is terse: it should state explicitly that n is a positive integer and that the sets S_0 \ S_r are relative complements; the current wording leaves the reader to guess the intended meaning for n=1 and for empty intersections.
  2. [Section 3, P(F) definition] The penalty P(F) is written with notation such as diag(M_ff) and M_ff - M_ff∘M; the dimensions and the use of the Hadamard product are clear only in context, so a brief sentence defining each term would improve readability.
  3. [Algorithm 1, Step 11] The truncation step enforces only the box constraint; the orthogonality constraints and the normalization (2) are not enforced by projection during the iterations. The text should state explicitly how the algorithm output is related to the constrained MLE given Theorem 3 and Proposition 1, rather than leaving the reader to infer this.
  4. [Abstract and Section 3 summary] The claim that the Lagrangian-type formulation 'coincides exactly' with the original problem omits the sign indeterminacy caveat that is correctly handled later in the paper; adding a short qualification would avoid overstatement.
  5. [Section 7] The PISA analysis assumes that booklet assignment is conditionally ignorable, but the model and likelihood used for the observed-response data are described in only one sentence; a fuller statement of the assumed missing-data mechanism and its justification would help practitioners assess the analysis.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the identifiability, estimation, and efficiency results are derived from stated assumptions and benchmarked against external oracle and Cramér-Rao quantities.

full rationale

The paper's central claims do not reduce to their inputs by construction. Theorem 1 proposes a combinatorial M-Q Condition and proves it necessary and sufficient for structural identifiability; although πM(AQ) is a closure system motivated by distinguishability, it is not defined in terms of Definition 1, and the paper verifies the condition in Examples 5–6 by independent algebraic arguments, such as the explicit rotation in Example 5. The Lagrangian reformulation in Section 3 is not a definitional equivalence: Theorem 3 has to prove that the global minimizer of Lν = L + νP satisfies the original constraints exactly, and the text explicitly identifies the difficulty that 'the first step only guarantees that P(˜F) is small, but not necessarily zero.' The proof is deferred to the supplementary material, which is a completeness risk rather than circularity: 'small P' is not 'zero P' by definition. The rates in Theorem 4 are compared with oracle rates from known-F or known-Λ subproblems, and the asymptotic variances in Theorem 5 are compared with the Cramér-Rao lower bounds, both external benchmarks not fitted in the paper. The choice ν=0.5 is an implementation constant asserted insensitive, not calibrated to data. Self-citations to Cui and Xu (2026a, 2026b) are contextual: the paper states that Cui and Xu (2026b) 'does not cover the present setting,' and Cui and Xu (2026a) is cited only for the general nonconvexity challenge, not as a substitute for Theorems 1–5. I therefore find no load-bearing circular step and score 0.

Assumptions & free parameters 3 free parameters · 9 assumptions · 3 invented entities

The theory rests on standard regularity conditions (Assumptions 1-2), the model-design assumption of the M-Q Condition, and the scaling condition J* proportional to J. No constants are fitted to data: the Lagrangian multiplier nu and truncation bound M are implementation choices with stated ranges, and the paper claims insensitivity to nu. The main invented constructs are the M-extended pi-system and the penalty P(F); both are internally justified by theorems and validated in examples.

free parameters (3)
  • Lagrangian multiplier nu = 0.5 (implementation)
    Algorithm 1 sets nu=0.5; theory requires only nu>nu0>0, and the paper asserts the results are insensitive to this value. It is a hand-chosen tuning constant, not fitted to data.
  • Truncation bound M = not specified
    The feasible set (7) and the truncation operator in Algorithm 1 need a constant M 'selected large enough'; no concrete value or data-driven rule is given, and the equivalence in Theorem 3 depends on it.
  • Gradient step-size eta = small constant satisfying eta J*/J <= 1/(2 gamma_delta)
    Proposition 1 fixes the step size through the curvature constant; in practice eta_f=eta/J, eta_lambda=eta_beta=eta/N are chosen small, but no default value is reported.
assumptions (9)
  • domain assumption The number of latent dimensions D is known and fixed.
    Stated in Section 2.1; selection of D is explicitly left to future work in Section 8.
  • domain assumption The response link functions g_ij are known and correctly specified.
    The log-likelihood is built from g_ij; the conclusion identifies misspecification as an open problem.
  • domain assumption The M-Q Condition holds for the pair (M, Q) under study.
    Theorem 1 makes it necessary for identifiability, and Theorems 2-5 assume it; practitioners must check it for their block design.
  • domain assumption Full column rank of each block loading matrix: rank(Lambda_k)=|A_k| for all k.
    Imposed in Section 2.1 to fix the factor-loading scale within blocks.
  • domain assumption Assumption 1: bounded true parameters; positive definite Sigma_f and Sigma_lambda; distinct eigenvalues of Sigma_f Sigma_lambda; per-block Gram matrices bounded away from zero.
    Standard factor model regularity conditions (Stock & Watson 2002, Bai 2003), used throughout Sections 3-5.
  • standard math Assumption 2: third-order smoothness, bounded second and third derivatives, and sub-exponential first derivative of the log-likelihood.
    Derivative concentration conditions used in the Hessian analysis; satisfied by logistic, probit, linear, and Poisson links.
  • domain assumption Assumption 3: J* is proportional to J and the growth condition log(N+J)((epsilon_J log N + epsilon_N log J)/(N and J))^{1/2} -> 0.
    Ensures strong convexity and negligibility of higher-order terms; restricts the asymptotic regimes covered by the theorems.
  • standard math Conditional independence of responses given latent variables.
    Assumed in Section 2.1 following Anderson (2003); needed for the likelihood factorization and the Fisher information formulas.
  • standard math Correctly specified parametric model satisfies E[l'^2] = -E[l''] under regularity conditions.
    Used in Theorem 5 to identify the sandwich covariance with the oracle Cramer-Rao bound.
invented entities (3)
  • M-extended pi-system pi_M(A) independent evidence
    purpose: Closure system over latent index sets used to define the M-Q Condition and decide structural identifiability for arbitrary block structure Q and orthogonality constraint M.
    Theorem 1 proves identifiability iff every singleton lies in pi_M(A_Q); Examples 5-6 show the condition toggles exactly with identifiability, and it subsumes known special-case identifiability results.
  • Quadratic penalty P(F) scaled by NJ/2 independent evidence
    purpose: Reformulates the normalization and partial orthogonality constraints as a penalty whose global minimizer coincides with the constrained maximum likelihood estimator.
    Theorem 3 establishes exact equivalence to the constrained MLE; the simulation coverage results and the nu-insensitivity statement support that the penalty does not bias the optimum.
  • Smallest non-removable block size J* independent evidence
    purpose: A summary statistic of the block design that controls the Hessian curvature lower bound and the convergence rate of the algorithm.
    J* is defined in (10) and enters Theorem 2 and Proposition 1 as the curvature scale; it is computable from Q and M and is unrelated to the observed data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Statistical analysis of block structured latent variable models." pith.science (2026). https://pith.science/paper/BO4FHVPX

@misc{pith2026260810328,
  author       = {Pith},
  title        = {Pith review of: Statistical analysis of block structured latent variable models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BO4FHVPX}},
  note         = {Machine review of arXiv:2608.10328}
}
read the original abstract

This paper studies block structured latent variable models, in which observed variables are grouped into distinct blocks based on their relationships with the underlying latent variables. These block structures are prevalent in various fields such as psychology, education, economics, and genetics. Despite their widespread applications, the fundamental statistical properties of these models remain largely unexplored. In this work, we present a comprehensive statistical analysis of the block structured latent variable models. In particular, we first derive conditions for model identifiability across various block designs. Furthermore, we investigate the maximum likelihood estimation under these identifiability constraints. To accommodate these intricate constraints associated with various block configurations, we introduce a Lagrangian-type formulation for the constrained nonconvex optimization problem and show that its optimum coincides with that of the original problem. This formulation serves as a critical tool for understanding the behavior of the constrained estimator under various block structures. Building on that, we establish sharp non-asymptotic error bounds and asymptotic distributions of the constrained maximum likelihood estimator. We also propose a computational framework to obtain the estimator and establish theoretical properties for the algorithm output. Our theoretical findings are validated through simulation studies and empirical data analyses.

Figures

Figures reproduced from arXiv: 2608.10328 by the authors.

Figure 1
Figure 1. Four examples of block structure designs in (a)–(d), with graphical representations shown in the upper panel and the corresponding matrix Q⊺ in the lower panel. In each graphical representation, dashed boxes denote blocks, and the nodes within them correspond to item response variables. Directed edges from latent factors to response variables indicate influence (absence of an edge implies independence). Undirected d… view at source ↗
Figure 2
Figure 2. Empirical coverage rates of factor parameters in the low-correlation case ( [PITH_FULL_IMAGE:figures/full_fig_p024_2.png] view at source ↗
Figure 3
Figure 3. Empirical coverage rates of loading parameters in the low-correlation case ( [PITH_FULL_IMAGE:figures/full_fig_p024_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Confidence intervals for the loading parameters on the general factor. Intervals in [PITH_FULL_IMAGE:figures/full_fig_p026_4.png]
Figure 5
Figure 5. Figure 5: Confidence intervals for the loading parameters on the three domain-specific [PITH_FULL_IMAGE:figures/full_fig_p026_5.png]
Figure 6
Figure 6. Figure 6: presents the joint density contours of the estimated latent abilities. The general, mathematics, and reading factors exhibit approximately normal distributions, whereas the science factor shows two distinct density peaks, suggesting a non-unimodal distribution in this …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

162 extracted references · 54 canonical work pages

  1. [1]

    Journal of Econometrics , volume=

    Maximum likelihood estimation and inference for high dimensional generalized factor models with application to factor-augmented regressions , author=. Journal of Econometrics , volume=. 2022 , publisher=

    2022

  2. [2]

    The Annals of Statistics , pages=

    The asymptotic normal distribution of estimators in factor analysis under general conditions , author=. The Annals of Statistics , pages=. 1988 , publisher=

    1988

  3. [3]

    The Annals of Statistics , volume=

    Partial identifiability of restricted latent class models , author=. The Annals of Statistics , volume=. 2020 , publisher=

    2020

  4. [4]

    The Annals of Statistics , volume=

    Bridging factor and sparse models , author=. The Annals of Statistics , volume=. 2023 , publisher=

    2023

  5. [5]

    2012 , publisher=

    Real and functional analysis , author=. 2012 , publisher=

    2012

  6. [6]

    Bridging convex and nonconvex optimization in robust

    Chen, Yuxin and Fan, Jianqing and Ma, Cong and Yan, Yuling , journal=. Bridging convex and nonconvex optimization in robust

  7. [7]

    The Annals of Statistics , volume=

    Noisy matrix decomposition via convex relaxation: Optimal rates in high dimensions , author=. The Annals of Statistics , volume=. 2012 , publisher=

    2012

  8. [8]

    1989 , publisher=

    Structural equations with latent variables , author=. 1989 , publisher=

    1989

Show all 162 references
  1. [9]

    , author=

    Evaluating bifactor models: Calculating and interpreting statistical indices. , author=. Psychological Methods , volume=. 2016 , publisher=

    2016

  2. [10]

    The Annals of Statistics , number =

    Jushan Bai and Kunpeng Li , title =. The Annals of Statistics , number =

  3. [11]

    Nature Methods , volume=

    Comparing the performance of biomedical clustering methods , author=. Nature Methods , volume=. 2015 , publisher=

    2015

  4. [12]

    Gu, Yuqi and Xu, Gongjun , journal=. A joint. 2023 , publisher=

    2023

  5. [13]

    The American Statistician , volume=

    The mean value theorem and Taylor’s expansion in statistics , author=. The American Statistician , volume=. 2013 , publisher=

    2013

  6. [14]

    arXiv preprint arXiv:2409.00679 , year=

    Exact Exploratory Bi-factor Analysis: A Constraint-based Optimisation Approach , author=. arXiv preprint arXiv:2409.00679 , year=

    arXiv

  7. [15]

    Psychometrika , volume=

    Exploratory bi-factor analysis: The oblique case , author=. Psychometrika , volume=. 2012 , publisher=

    2012

  8. [16]

    Statistica Sinica , volume=

    Identifiability of bifactor models , author=. Statistica Sinica , volume=. 2021 , publisher=

    2021

  9. [17]

    Personality Psychology in Europe , volume=

    A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models , author=. Personality Psychology in Europe , volume=. 1999 , publisher=

    1999

  10. [18]

    A useful variant of the

    Yu, Yi and Wang, Tengyao and Samworth, Richard J , journal=. A useful variant of the. 2015 , publisher=

    2015

  11. [19]

    2000 , publisher=

    Asymptotic statistics , author=. 2000 , publisher=

    2000

  12. [20]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume =

    Rohe, Karl and Zeng, Muzhe , title = ". Journal of the Royal Statistical Society Series B: Statistical Methodology , volume =. 2023 , month =

    2023

  13. [21]

    arXiv preprint arXiv:2310.10545 , year=

    Optimal vintage factor analysis with deflation varimax , author=. arXiv preprint arXiv:2310.10545 , year=

    arXiv

  14. [22]

    Journal of Econometrics , volume=

    Principal components estimation and identification of static factors , author=. Journal of Econometrics , volume=. 2013 , publisher=

    2013

  15. [23]

    Psychometrika , volume=

    Rotation for simple loadings , author=. Psychometrika , volume=. 1966 , publisher=

    1966

  16. [24]

    Psychometrika , volume=

    Rotation to simple loadings using component loss functions: The orthogonal case , author=. Psychometrika , volume=. 2004 , publisher=

    2004

  17. [25]

    2013 , publisher=

    Introductory lectures on convex optimization: A basic course , author=. 2013 , publisher=

    2013

  18. [26]

    Psychometrika , volume=

    Rotation to simple loadings using component loss functions: The oblique case , author=. Psychometrika , volume=. 2006 , publisher=

    2006

  19. [27]

    Rotation to Sparse Loadings Using

    Xinyi Liu and Gabriel Wallin and Yunxiao Chen and Irini Moustaki , journal=. Rotation to Sparse Loadings Using. 2023 , volume=

    2023

  20. [28]

    arXiv preprint arXiv:1011.3027 , year=

    Introduction to the non-asymptotic analysis of random matrices , author=. arXiv preprint arXiv:1011.3027 , year=

    arXiv

  21. [29]

    Annals of Probability , volume=

    Sharp nonasymptotic bounds on the norm of random matrices with independent entries , author=. Annals of Probability , volume=. 2016 , publisher=

    2016

  22. [30]

    Psychometrika , volume=

    The varimax criterion for analytic rotation in factor analysis , author=. Psychometrika , volume=. 1958 , publisher=

    1958

  23. [31]

    Journal of the American Statistical Association , pages=

    Inferring covariance structure from multiple data sources via subspace factor analysis , author=. Journal of the American Statistical Association , pages=. 2024 , publisher=

    2024

  24. [32]

    Journal of the Royal Statistical Society: Series B (Methodological) , volume=

    Parameter orthogonality and approximate conditional inference , author=. Journal of the Royal Statistical Society: Series B (Methodological) , volume=. 1987 , publisher=

    1987

  25. [33]

    Journal of Econometrics , volume=

    The incidental parameter problem since 1948 , author=. Journal of Econometrics , volume=. 2000 , publisher=

    1948

  26. [34]

    Measuring thirty facets of the Five Factor Model with a 120-item public domain inventory: Development of the

    Johnson, John A , journal=. Measuring thirty facets of the Five Factor Model with a 120-item public domain inventory: Development of the. 2014 , publisher=

    2014

  27. [35]

    2023 , publisher=

    PISA 2022 assessment and analytical framework , author=. 2023 , publisher=

    2022

  28. [36]

    Journal of Machine Learning Research , year=

    A max-norm constrained minimization approach to 1-bit matrix completion , author=. Journal of Machine Learning Research , year=

  29. [37]

    Journal of the American statistical association , volume=

    Forecasting using principal components from a large number of predictors , author=. Journal of the American statistical association , volume=. 2002 , publisher=

    2002

  30. [38]

    From simple structure to sparse components:

    Trendafilov, Nickolay T , journal=. From simple structure to sparse components:. 2014 , publisher=

    2014

  31. [39]

    arXiv preprint arXiv:1412.5647 , year=

    Nonlinear panel models with interactive effects , author=. arXiv preprint arXiv:1412.5647 , year=

    arXiv

  32. [40]

    Journal of the American Statistical Association , volume=

    Structured latent factor analysis for large-scale data: Identifiability, estimability, and their implications , author=. Journal of the American Statistical Association , volume=. 2020 , publisher=

    2020

  33. [41]

    Handbook of the Fundamentals of Financial Decision Making: Part I , pages=

    The arbitrage theory of capital asset pricing , author=. Handbook of the Fundamentals of Financial Decision Making: Part I , pages=. 2013 , publisher=

    2013

  34. [42]

    Econometrica , volume=

    Determining the number of factors in approximate factor models , author=. Econometrica , volume=. 2002 , publisher=

    2002

  35. [43]

    Behavior Genetics , volume=

    Exploratory factor analysis in behavior genetics research: Factor recovery with small sample sizes , author=. Behavior Genetics , volume=. 2002 , publisher=

    2002

  36. [44]

    , author=

    Sample size in factor analysis. , author=. Psychological Methods , volume=. 1999 , publisher=

    1999

  37. [45]

    Journal of the Royal Statistical Society

    Factor analysis as a statistical method , author=. Journal of the Royal Statistical Society. Series D (The Statistician) , volume=. 1962 , publisher=

    1962

  38. [46]

    Econometrica , volume=

    Inferential theory for factor models of large dimensions , author=. Econometrica , volume=. 2003 , publisher=

    2003

  39. [47]

    Psychometrika , volume=

    Joint maximum likelihood estimation for high-dimensional exploratory item factor analysis , author=. Psychometrika , volume=. 2019 , publisher=

    2019

  40. [48]

    2009 , publisher=

    Confirmatory factor analysis , author=. 2009 , publisher=

    2009

  41. [49]

    The Annals of Applied Statistics, to appear , year=

    Statistical Inference for Covariate-Adjusted and Interpretable Generalized Factor Model with Application to Testing Fairness , author=. The Annals of Applied Statistics, to appear , year=

  42. [50]

    Bi-cross-validation for factor analysis , author=

  43. [51]

    arXiv preprint arXiv:2505.09043 , year=

    Exploratory Hierarchical Factor Analysis with an Application to Psychological Measurement , author=. arXiv preprint arXiv:2505.09043 , year=

  44. [52]

    Journal of the American Statistical Association , number=

    Identifiability and inference for generalized latent factor models , author=. Journal of the American Statistical Association , number=. 2026 , publisher=

    2026

  45. [53]

    IEEE Transactions on knowledge and data engineering , volume=

    Cluster analysis for gene expression data: a survey , author=. IEEE Transactions on knowledge and data engineering , volume=. 2004 , publisher=

    2004

  46. [54]

    2004 , publisher=

    Generalized latent variable modeling: Multilevel, longitudinal, and structural equation models , author=. 2004 , publisher=

    2004

  47. [55]

    2013 , publisher=

    Item response theory , author=. 2013 , publisher=

    2013

  48. [56]

    Journal of personality assessment , volume=

    Mini-Markers: A brief version of Goldberg's unipolar Big-Five markers , author=. Journal of personality assessment , volume=. 1994 , publisher=

    1994

  49. [57]

    Journal of Machine Learning Research , volume=

    Statistical inference for noisy incomplete binary matrix , author=. Journal of Machine Learning Research , volume=

  50. [58]

    Journal of the American Statistical Association , volume=

    Clustering huge number of financial time series: A panel data approach with high-dimensional predictors and factor structures , author=. Journal of the American Statistical Association , volume=. 2017 , publisher=

    2017

  51. [59]

    Econometrica , volume=

    Confidence intervals for diffusion index forecasts and inference for factor-augmented regressions , author=. Econometrica , volume=. 2006 , publisher=

    2006

  52. [60]

    2008 , publisher=

    Econometric analysis of count data , author=. 2008 , publisher=

    2008

  53. [61]

    2007 , publisher=

    Linear algebra and its applications , author=. 2007 , publisher=

    2007

  54. [62]

    Proceedings of the Berkeley Symposium on Mathematical Statistics and Probability , pages=

    STATISTICAL INFERENCE IN , author=. Proceedings of the Berkeley Symposium on Mathematical Statistics and Probability , pages=. 1956 , organization=

    1956

  55. [63]

    The Annals of Statistics , pages=

    Factor modeling for high-dimensional time series: inference for the number of factors , author=. The Annals of Statistics , pages=. 2012 , publisher=

    2012

  56. [64]

    The Annals of Statistics , pages=

    Adaptive estimation in structured factor models with applications to overlapping clustering , author=. The Annals of Statistics , pages=

  57. [65]

    American Educational Research Journal , volume=

    Enhancing the validity and usefulness of large-scale educational assessments: IV NELS: 88 science achievement to 12th grade , author=. American Educational Research Journal , volume=. 1997 , publisher=

    1997

  58. [66]

    The Annals of Mathematical Statistics , volume=

    Asymptotic properties of non-linear least squares estimators , author=. The Annals of Mathematical Statistics , volume=. 1969 , publisher=

    1969

  59. [67]

    The Annals of Statistics , volume=

    Asymptotic theory of nonlinear least squares estimation , author=. The Annals of Statistics , volume=. 1981 , publisher=

    1981

  60. [68]

    American Psychological Association , year=

    Exploratory and Confirmatory Factor Analysis: Understanding Concepts and Applications , author=. American Psychological Association , year=

  61. [69]

    2003 , publisher=

    Statistical analysis of management data , author=. 2003 , publisher=

    2003

  62. [70]

    Educational Research Review , volume=

    Large-scale student assessment studies measure the results of processes of knowledge acquisition: Evidence in support of the distinction between intelligence and student achievement , author=. Educational Research Review , volume=. 2009 , publisher=

    2009

  63. [71]

    Geometrical methods for non-negative

    Plumbley, Mark D , journal=. Geometrical methods for non-negative. 2005 , publisher=

    2005

  64. [72]

    2009 , publisher=

    Multidimensional Item Response Theory , author=. 2009 , publisher=

    2009

  65. [73]

    The Annals of Probability , pages=

    Convergence in distribution of conditional expectations , author=. The Annals of Probability , pages=. 1994 , publisher=

    1994

  66. [74]

    2017 , publisher=

    Probability and measure , author=. 2017 , publisher=

    2017

  67. [75]

    , author=

    Confirmatory factor analysis and item response theory: two approaches for exploring measurement invariance. , author=. Psychological bulletin , volume=. 1993 , publisher=

    1993

  68. [76]

    Psychometrika , volume=

    Full-information item bi-factor analysis , author=. Psychometrika , volume=. 1992 , publisher=

    1992

  69. [77]

    psychometrika , volume=

    Semiparametric factor analysis for item-level response time data , author=. psychometrika , volume=. 2022 , publisher=

    2022

  70. [78]

    Journal of Econometrics , volume=

    Deep learning based residuals in non-linear factor models: Precision matrix estimation of returns with low signal-to-noise ratio , author=. Journal of Econometrics , volume=. 2025 , publisher=

    2025

  71. [79]

    2002 , publisher=

    Introduction to nonparametric item response theory , author=. 2002 , publisher=

    2002

  72. [80]

    arXiv preprint arXiv:2605.05446 , year=

    Convexity in disguise: A theoretical framework for nonconvex low-rank matrix estimation , author=. arXiv preprint arXiv:2605.05446 , year=

    arXiv

  73. [81]

    2014 , publisher=

    Constrained optimization and Lagrange multiplier methods , author=. 2014 , publisher=

    2014

  74. [82]

    , author=

    Generalized full-information item bifactor analysis. , author=. Psychological methods , volume=. 2011 , publisher=

    2011

  75. [83]

    The Annals of Statistics , volume=

    Half-trek criterion for identifiability of latent variable models , author=. The Annals of Statistics , volume=. 2022 , publisher=

    2022

  76. [84]

    arXiv preprint arXiv:2312.14762 , year=

    Algebraic Sparse Factor Analysis , author=. arXiv preprint arXiv:2312.14762 , year=

    arXiv

  77. [85]

    Psychometrika , volume=

    Marginal maximum likelihood estimation of item parameters: Application of an EM algorithm , author=. Psychometrika , volume=. 1981 , publisher=

    1981

  78. [86]

    Journal of Multivariate Analysis , volume=

    Angle-based joint and individual variation explained , author=. Journal of Multivariate Analysis , volume=. 2018 , publisher=

    2018

  79. [87]

    2013 , publisher=

    An introduction to latent variable models , author=. 2013 , publisher=

    2013

  80. [88]

    2004 , publisher=

    Latent variable models: An introduction to factor, path, and structural equation analysis , author=. 2004 , publisher=

    2004

  81. [89]

    Journal of the American Statistical Association , volume=

    Integrating multisource block-wise missing data in model selection , author=. Journal of the American Statistical Association , volume=. 2021 , publisher=

    2021

  82. [90]

    Statistica Sinica , volume=

    Imputed factor regression for high-dimensional block-wise missing data , author=. Statistica Sinica , volume=. 2020 , publisher=

    2020

  83. [91]

    Biostatistics , volume=

    Generalized integrative principal component analysis for multi-type data with block-wise missing structure , author=. Biostatistics , volume=. 2020 , publisher=

    2020

  84. [92]

    Journal of Machine Learning Research , volume=

    Multi-source learning via completion of block-wise overlapping noisy matrices , author=. Journal of Machine Learning Research , volume=

  85. [93]

    2011 , publisher=

    Latent variable models and factor analysis: A unified approach , author=. 2011 , publisher=

    2011

  86. [94]

    IEEE Transactions on Information Theory , volume=

    Matrix completion from a few entries , author=. IEEE Transactions on Information Theory , volume=. 2010 , publisher=

    2010

  87. [95]

    The Annals of Statistics , volume=

    Inference for heteroskedastic PCA with missing data , author=. The Annals of Statistics , volume=. 2024 , publisher=

    2024

  88. [96]

    Journal of the American Statistical Association , volume=

    High-dimensional sparse factor modeling: applications in gene expression genomics , author=. Journal of the American Statistical Association , volume=. 2008 , publisher=

    2008

  89. [97]

    How common are common return factors across the

    Goyal, Amit and P. How common are common return factors across the. Journal of Financial Economics , volume=. 2008 , publisher=

    2008

  90. [98]

    Journal of Financial Economics , volume=

    Multifactor asset pricing with a large number of observable risk factors and unobservable common and group-specific factors , author=. Journal of Financial Economics , volume=. 2015 , publisher=

    2015

  91. [99]

    The Annals of Statistics , volume=

    Maximum likelihood estimates in exponential response models , author=. The Annals of Statistics , volume=. 1977 , publisher=

    1977

  92. [100]

    large p , small n

    Bayesian factor regression models in the “large p , small n ” paradigm , author=. Bayesian statistics , volume=

  93. [101]

    2013 , publisher=

    Item response theory: Principles and applications , author=. 2013 , publisher=

    2013

  94. [102]

    Nature Methods , volume=

    A new Bayesian factor analysis method improves detection of genes and biological processes affected by perturbations in single-cell CRISPR screening , author=. Nature Methods , volume=. 2023 , publisher=

    2023

  95. [103]

    arXiv preprint arXiv:2110.10804 , year=

    Identifiable deep generative models via sparse decoding , author=. arXiv preprint arXiv:2110.10804 , year=

    arXiv

  96. [104]

    Structural Equation Modeling: A Multidisciplinary Journal , volume=

    On desirability of parsimony in structural equation model selection , author=. Structural Equation Modeling: A Multidisciplinary Journal , volume=. 1999 , publisher=

    1999

  97. [105]

    The journal of experimental education , volume=

    Assessing goodness of fit: Is parsimony always desirable? , author=. The journal of experimental education , volume=. 1996 , publisher=

    1996

  98. [106]

    Molecular Biology and Evolution , volume=

    Testing for associations between loci and environmental gradients using latent factor mixed models , author=. Molecular Biology and Evolution , volume=. 2013 , publisher=

    2013

  99. [107]

    Annual Review of Clinical Psychology , volume=

    Exploratory structural equation modeling: An integration of the best features of exploratory and confirmatory factor analysis , author=. Annual Review of Clinical Psychology , volume=. 2014 , publisher=

    2014

  100. [108]

    2015 , publisher=

    Confirmatory factor analysis for applied research , author=. 2015 , publisher=

    2015

  101. [109]

    IEEE Transactions on Signal Processing , volume=

    Multi-channel factor analysis with common and unique factors , author=. IEEE Transactions on Signal Processing , volume=. 2019 , publisher=

    2019

  102. [110]

    Nature Communications , volume=

    Benchmarking joint multi-omics dimensionality reduction approaches for the study of cancer , author=. Nature Communications , volume=. 2021 , publisher=

    2021

  103. [111]

    Biometrika , volume=

    Determining the number of factors in high-dimensional generalized latent factor models , author=. Biometrika , volume=. 2022 , publisher=

    2022

  104. [112]

    Briefings in Bioinformatics , volume=

    A review on machine learning principles for multi-view biological data integration , author=. Briefings in Bioinformatics , volume=. 2018 , publisher=

    2018

  105. [113]

    arXiv preprint arXiv:2501.01414 , year=

    Deep discrete encoders: Identifiable deep generative models for rich data with discrete latent layers , author=. arXiv preprint arXiv:2501.01414 , year=

    arXiv

  106. [114]

    arXiv preprint arXiv:2410.12108 , year=

    A general latent embedding approach for modeling non-uniform high-dimensional sparse hypergraphs with multiplicity , author=. arXiv preprint arXiv:2410.12108 , year=

  107. [115]

    arXiv preprint arXiv:2312.06605 , year=

    Statistical inference on latent space models for network data , author=. arXiv preprint arXiv:2312.06605 , year=

    arXiv

  108. [116]

    Proceedings of the 24th Annual Conference on Learning Theory , pages=

    Concentration-based guarantees for low-rank matrix reconstruction , author=. Proceedings of the 24th Annual Conference on Learning Theory , pages=. 2011 , organization=

    2011

  109. [117]

    Biometrics , volume=

    Multi-study factor analysis , author=. Biometrics , volume=. 2019 , publisher=

    2019

  110. [118]

    1976 , publisher=

    Modern factor analysis , author=. 1976 , publisher=

    1976

  111. [119]

    Psychometrika , volume=

    A two-tier full-information item factor analysis model with applications , author=. Psychometrika , volume=. 2010 , publisher=

    2010

  112. [120]

    arXiv preprint arXiv:1509.03025 , year=

    Fast low-rank estimation by projected gradient descent: General statistical and algorithmic guarantees , author=. arXiv preprint arXiv:1509.03025 , year=

    arXiv

  113. [121]

    Mathematical Programming , volume=

    Local minima and convergence in low-rank semidefinite programming , author=. Mathematical Programming , volume=. 2005 , publisher=

    2005

  114. [122]

    The Annals of Statistics , volume=

    Model assisted variable clustering: Minimax-optimal recovery and algorithms , author=. The Annals of Statistics , volume=

  115. [123]

    Higher-order factors of the

    Digman, John M , journal=. Higher-order factors of the. 1997 , publisher=

    1997

  116. [124]

    The revised

    Costa, Paul T and McCrae, Robert R , journal=. The revised

  117. [125]

    , author=

    Validation of the five-factor model of personality across instruments and observers. , author=. Journal of personality and social psychology , volume=. 1987 , publisher=

    1987

  118. [126]

    Journal of personality , volume=

    An introduction to the five-factor model and its applications , author=. Journal of personality , volume=. 1992 , publisher=

    1992

  119. [127]

    Personality and Individual Differences , volume=

    The five factor model and impulsivity: Using a structural model of personality to understand impulsivity , author=. Personality and Individual Differences , volume=. 2001 , publisher=

    2001

  120. [128]

    Journal of Educational Change , volume=

    Can competencies assessed by PISA be considered the fundamental school knowledge 15-year-olds should possess? , author=. Journal of Educational Change , volume=. 2007 , publisher=

    2007

  121. [129]

    How much do students’ scores in

    Pokropek, Artur and Marks, Gary N and Borgonovi, Francesca , journal=. How much do students’ scores in. 2022 , publisher=

    2022

  122. [130]

    Brunner, Martin , journal=. No. 2008 , publisher=

    2008

  123. [131]

    Multivariate Behavioral Research , volume=

    General and specific abilities as predictors of school achievement , author=. Multivariate Behavioral Research , volume=. 1993 , publisher=

    1993

  124. [132]

    Psychometrika , volume=

    A general approach to confirmatory maximum likelihood factor analysis , author=. Psychometrika , volume=. 1969 , publisher=

    1969

  125. [133]

    2003 , publisher=

    An introduction to multivariate statistical analysis , author=. 2003 , publisher=

    2003

  126. [134]

    Journal of Econometrics , volume=

    Sufficient forecasting using factor models , author=. Journal of Econometrics , volume=. 2017 , publisher=

    2017

  127. [135]

    IEEE transactions on information theory , volume=

    The power of convex relaxation: Near-optimal matrix completion , author=. IEEE transactions on information theory , volume=. 2010 , publisher=

    2010

  128. [136]

    Nuclear-norm penalization and optimal rates for noisy low-rank matrix completion , author=

  129. [137]

    Journal of Machine Learning Research , volume=

    Universal latent space model fitting for large networks with edge covariates , author=. Journal of Machine Learning Research , volume=

  130. [138]

    Information and Inference: A Journal of the IMA , volume=

    1-bit matrix completion , author=. Information and Inference: A Journal of the IMA , volume=. 2014 , publisher=

    2014

  131. [139]

    , author=

    Multiple factor analysis. , author=. 1947 , publisher=

    1947

  132. [140]

    1987 , publisher=

    Multivariate exploratory data analysis: A perspective on exploratory factor analysis , author=. 1987 , publisher=

    1987

  133. [141]

    Multivariate Behavioral Research , volume=

    An overview of analytic rotation in exploratory factor analysis , author=. Multivariate Behavioral Research , volume=. 2001 , publisher=

    2001

  134. [142]

    Psychometrika , volume=

    The bi-factor method , author=. Psychometrika , volume=. 1937 , publisher=

    1937

  135. [143]

    Journal of the American Statistical Association , volume=

    Model-based clustering, discriminant analysis, and density estimation , author=. Journal of the American Statistical Association , volume=. 2002 , publisher=

    2002

  136. [144]

    Structural Equation Modeling: A Multidisciplinary Journal , volume=

    Multiple-group factor analysis alignment , author=. Structural Equation Modeling: A Multidisciplinary Journal , volume=. 2014 , publisher=

    2014

  137. [145]

    Artificial Intelligence and Statistics , pages=

    Bayesian group factor analysis , author=. Artificial Intelligence and Statistics , pages=. 2012 , organization=

    2012

  138. [146]

    Joint and individual variation explained (

    Lock, Eric F and Hoadley, Katherine A and Marron, James Stephen and Nobel, Andrew B , journal=. Joint and individual variation explained (. 2013 , publisher=

    2013

  139. [147]

    Econometrica , volume=

    Inference in group factor models with an application to mixed-frequency data , author=. Econometrica , volume=. 2019 , publisher=

    2019

  140. [148]

    Econometric Reviews , volume=

    Identification of the linear factor model , author=. Econometric Reviews , volume=. 2020 , publisher=

    2020

  141. [149]

    IEEE Transactions on Signal Processing , volume=

    Nonconvex optimization meets low-rank matrix factorization: An overview , author=. IEEE Transactions on Signal Processing , volume=. 2019 , publisher=

    2019

  142. [150]

    Mathematical programming , volume=

    A nonlinear programming algorithm for solving semidefinite programs via low-rank factorization , author=. Mathematical programming , volume=. 2003 , publisher=

    2003

  143. [151]

    SIAM review , volume=

    On the perturbation of pseudo-inverses, projections and linear least squares problems , author=. SIAM review , volume=. 1977 , publisher=

    1977

  144. [152]

    SIAM Journal on Optimization , volume=

    Noisy matrix completion: Understanding statistical guarantees for convex relaxation via nonconvex optimization , author=. SIAM Journal on Optimization , volume=. 2020 , publisher=

    2020

  145. [153]

    International Conference on Machine Learning , pages=

    Implicit regularization in nonconvex statistical estimation: Gradient descent converges linearly for phase retrieval and matrix completion , author=. International Conference on Machine Learning , pages=. 2018 , organization=

    2018

  146. [154]

    IEEE Transactions on Information Theory , volume=

    Guaranteed matrix completion via non-convex factorization , author=. IEEE Transactions on Information Theory , volume=. 2016 , publisher=

    2016

  147. [155]

    Bulletin of the American Mathematical Society , year=

    Variational methods for the solution of problems of equilibrium and vibrations , author=. Bulletin of the American Mathematical Society , year=

  148. [156]

    1999 , publisher=

    Numerical optimization , author=. 1999 , publisher=

    1999

  149. [157]

    Journal of Financial Econometrics , volume=

    Bootstrap inference for group factor models , author=. Journal of Financial Econometrics , volume=. 2025 , publisher=

    2025

  150. [158]

    Journal of the American Statistical Association , volume=

    Are latent factor regression and sparse regression adequate? , author=. Journal of the American Statistical Association , volume=. 2024 , publisher=

    2024

  151. [159]

    Journal of the American Statistical Association , volume=

    Embracing the blessing of dimensionality in factor models , author=. Journal of the American Statistical Association , volume=. 2018 , publisher=

    2018

  152. [160]

    Journal of Econometrics , volume=

    Nonlinear factor models for network and panel data , author=. Journal of Econometrics , volume=. 2021 , publisher=

    2021

  153. [161]

    Individual and time effects in nonlinear panel models with large

    Fern. Individual and time effects in nonlinear panel models with large. Journal of Econometrics , volume=. 2016 , publisher=

    2016

  154. [162]

    Data integration via analysis of subspaces (

    Prothero, Jack and Jiang, Meilei and Hannig, Jan and Tran-Dinh, Quoc and Ackerman, Andrew and Marron, JS , journal=. Data integration via analysis of subspaces (. 2024 , publisher=

    2024

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.