Pith. sign in

REVIEW 3 major objections 7 minor 36 references

On improving the estimates of the sampling variances via Global-Local priors in Small Area Estimation

T0 review · 3 major / 7 minor · reviewed 2026-07-10 · glm-5.2

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

Pith's one-line read Shrinkage that adapts: Global-Local priors sharpen small-area variance estimates

desk verdict GL shrinkage priors for sampling variances in the Fay-Herriot model: solid theory, but simulations favor the proposed model's home turf read the letter →

arxiv 2607.08720 v1 pith:JB5G4GRG submitted 2026-07-09 stat.ME

Sirapat Watakajaturaphon , Jairo Fúquene-Patiño This is my paper
classification stat.ME
keywords modelestimatesestimationproposedsamplingsmallvariancesaddress
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 proposes a new Bayesian model for Small Area Estimation (SAE) that improves estimates of sampling variances — the uncertainty attached to survey-based estimates for small geographic or demographic domains. In the classical Fay-Herriot model, these variances are treated as known, but in practice they are themselves estimated from limited data and can be wildly unreliable. Practitioners smooth them using Generalized Variance Functions (GVFs), which fit a regression of log-variance on covariates like sample size. The authors argue that existing Bayesian approaches either do not shrink toward the GVF fit at all, or do so in a rigid way that cannot adapt to heterogeneity across areas. Their solution is to place a Global-Local (GL) shrinkage prior on each area's sampling variance: a global parameter controls overall shrinkage toward the GVF-smoothed value, while a local parameter lets each area deviate when its own data warrant it. They prove that the Beta Prime distribution with specific hyperparameters is the right choice for the local parameter because it produces a prior that both shrinks aggressively toward the GVF estimate and remains heavy-tailed enough to accommodate genuine outliers. The posterior mean of each variance becomes a data-driven weighted average of the raw variance estimate and the GVF-smoothed value, with weights that depend on both sample size and how well the GVF explains that area's variance. Simulation studies show the model yields lower squared error for both area means and variances, and two real applications — U.S. corn production and educational attainment in Colombian municipalities — confirm the gains.

What carries the argument

The model layers three components: (1) the Fay-Herriot area-level model linking direct survey estimates to covariates via a regression with random effects; (2) a Gamma sampling model for the direct variance estimates D_i given the true sampling variances σ²_i, with degrees of freedom ν_i = n_i − 1; and (3) an Inverse-Gamma prior for σ²_i whose shape and scale both involve a global parameter α and a local parameter ω_i, centered at the Exp-GVF exp(z_i^T η). The local parameter ω_i receives a Beta Prime prior BP(2, 1/2), chosen because it satisfies three properties: non-shrinkage at the origin (the prior density vanishes as σ²_i → 0), a heavy right tail (polynomial decay), and a spike at the G

What would settle it

If, in a simulation where the true variances deviate substantially from the GVF structure (e.g., a large fraction of areas have variance driven by an unobserved area-specific factor), the GL-BP model's adaptive shrinkage fails to pull back from the GVF target and produces higher squared error than the non-shrinking YC model, the core claim of adaptive superiority would be undermined. The simulation Cases 2–4 test versions of this scenario, but only with deviations that are mixtures of the GVF structure and point masses — not with a fundamentally different variance-generating mechanism.

Watch

Extended reading notes

Core claim

The central mechanism is a shrinkage factor that adapts area by area. Under the proposed model, the posterior mean of each sampling variance σ²_i takes the form of a weighted average between an observed variance (driven by the direct estimate D_i and the area mean θ_i) and the GVF-smoothed value exp(z_i^T η). The weight, called the posterior shrinkage factor κ_post_i, depends on the global parameter α, the local parameter ω_i, and the sample size n_i. When sample sizes are small and the GVF fits well, the factor approaches one and the variance estimate is pulled strongly toward the GVF value. When sample sizes are large or the GVF fits poorly, the factor approaches zero and the raw data are信

Load-bearing premise

The Gamma sampling model for direct variance estimates is theoretically justified only under simple random sampling, but the Colombian education application uses a complex survey design. The authors approximate the degrees of freedom using the Delta method, but the adequacy of this approximation for the Gamma model's tail behavior — and hence for the calibration of the shrinkage weights — is not validated.

Editorial extensions

If this is right

  • Official statistics agencies producing subnational estimates could adopt the GL-BP model to reduce reliance on unstable direct variance estimates, particularly for domains with very small sample sizes where current methods either over-smooth or fail to shrink.
  • The adaptive shrinkage framework could be extended to other survey-derived quantities beyond variances — for example, design effects or non-response rates — wherever a GVF-type smoothing model exists and the raw estimates are noisy.
  • The identifiability advantage of the GL-BP model over the STK2 model (which cannot separately identify an intercept in the GVF from a multiplicative scale parameter) suggests the proposed parameterization is more natural for problems where the GVF includes multiple covariates.
  • The Delta-method approximation for degrees of freedom under complex survey designs, used here for the Colombian education application, could be validated against design-based simulation studies to assess its adequacy for prevalence-type estimates.

Reading between the lines

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

  • The adaptive shrinkage mechanism implicitly assumes that the GVF covariates (here, log sample size and log projected households) are the correct carriers of cross-area information. If an important area-level covariate driving variance heterogeneity is omitted from the GVF, the model will shrink toward a misspecified target, and the local parameter may not fully compensate — the heavy tail mitigate
  • The choice of a Gamma(1, 1) prior for the global parameter α means the model starts with a weakly informative assumption about overall shrinkage strength. In applications where the GVF is known to be highly predictive, a more informative prior on α could accelerate convergence and improve finite-sample performance.
  • The framework could in principle be extended to multivariate small area problems where multiple correlated indicators are estimated simultaneously, with a joint Global-Local prior capturing cross-indicator as well as cross-area heterogeneity.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. This manuscript proposes a Bayesian Global-Local (GL) shrinkage prior for the sampling variances in the modified Fay-Herriot model for Small Area Estimation (SAE). The core idea is to place an Inverse-Gamma prior on each area's sampling variance σ²_i with shape and scale governed by a global parameter α and area-specific local parameter ω_i, centered at the Generalized Variance Function (GVF) estimate exp(z_i^T η). The authors study marginal prior properties under Beta Prime (BP) and Gamma local priors, derive the posterior shrinkage factor structure, prove posterior concentration of the shrinkage factor (Theorem 2.1), establish posterior propriety (Theorem 3.1), and develop adaptive MCMC algorithms. The proposed GL-BP model is compared against three existing Bayesian SAE models (YC, STK1, STK2) in simulation studies and two real applications (Corn production, Educational Attainment Index in Colombia).

Significance. The paper addresses a practically important problem in SAE: the instability of direct variance estimates and the need for adaptive shrinkage toward GVF-smoothed estimates. The theoretical contributions are substantive: Corollary 2.1 and Proposition 2.1 provide a principled basis for selecting the BP local prior hyperparameters (a=2, b=1/2) by verifying three desirable marginal prior properties; Theorem 2.1 gives posterior concentration of the shrinkage factor; and Theorem 3.1 establishes posterior propriety under mild conditions. The adaptive MCMC algorithm with variance tuning for the non-conjugate ξ_i and η updates is a practical computational contribution. The PEAI-AHS application involving 289 Colombian municipalities is a novel and policy-relevant dataset. The key methodological distinction from STK2 is that the shrinkage factor depends on both the global/local parameters and sample size, rather than sample size alone, enabling data-driven adaptation.

major comments (3)
  1. Section 4, simulation design: In all four simulation cases, the true σ²_i is generated as c·exp(η·log(n_i)) (possibly multiplied by random λ_i), which is exactly the functional form of the Exp-GVF in the GL-BP prior (Eq. 5). This means the GVF is correctly specified for GL-BP in every simulation scenario. The ASD improvements for σ²_i are dramatic (e.g., 0.134 vs. 0.672 for STK2 in Case 1, Table 2—a 5x reduction), but this advantage is partly mechanical: when the GVF is correctly specified, aggressive shrinkage toward it is naturally rewarded. While Cases 2–4 introduce heterogeneity via λ_i, the deviations are structured as point-mass-plus-Log-Normal noise rather than systematic GVF misspecification (e.g., σ²_i depending on an unmodeled covariate or a different functional form of n_i). The paper would be substantially strengthened by adding at least one simulation case where the GVF is系统
  2. Section 4, Table 2: The ASD improvements for θ_i are modest (e.g., 1.339 vs. 1.344 for STK2 in Case 1), while the improvements for σ²_i are dramatic (0.134 vs. 0.672). Since θ_i is the primary parameter of interest in SAE, the practical significance of the σ²_i improvements should be discussed more carefully. The authors should clarify whether the θ_i improvements, though consistent, are practically meaningful in typical SAE applications. The coverage probabilities for θ_i are comparable across models (94.0–95.1%), so the main benefit appears to be in variance estimation rather than mean estimation. This should be stated transparently.
  3. Section 5.2.1 and Appendix D: The Gamma sampling model D_i | σ²_i ~ Ga(ν_i/2, ν_i/(2σ²_i)) with ν_i = n_i − 1 is theoretically justified only under simple random sampling. For the PEAI-AHS application, which uses a complex DHS survey design, the authors approximate ν_i via the Delta method as ν̂_i = 2n_i y_i(1−y_i)/(1−2y_i)². However, the adequacy of this approximation for the Gamma model's tail behavior—which directly drives the shrinkage weights κ_post_i—is not validated. When y_i approaches 0.5, ν̂_i becomes very large (the denominator (1−2y_i)² → 0), which could produce extreme shrinkage behavior. The authors should discuss the sensitivity of their results to this approximation, perhaps by examining the distribution of ν̂_i values in the PEAI-AHS data or by a sensitivity analysis with alternative ν_i specifications.
minor comments (7)
  1. Section 2.2.1, Figure 1: The right panel showing tail behavior uses a very different x-axis range (up to σ²=1000) than the left panel (up to σ²=2). It would help to add a vertical reference line at σ²=1 (the Exp-GVF value) in both panels.
  2. Section 3.2, Algorithm 1, Step 11: The adaptive MH scheme updates the proposal variance every ℓ=200 iterations. It would be useful to report the final acceptance rates achieved in the simulation and real data applications to confirm the adaptive scheme is working as intended.
  3. Table 1: The MCMC column for STK2 says 'Gibbs and adaptive Metropolis-Hastings' but the description in Section 3.1 mentions that the original Sugasawa et al. (2017) used a fixed-variance MH. Clarify whether the adaptive MH for STK2 is your modification or the original approach.
  4. Section 5.1, Table 3: The DIC for YC is −10.34 while for GL-BP it is −17.63. Given that these models have different numbers of effective parameters, it would be informative to also report p_D (the effective number of parameters) alongside DIC.
  5. Section 5.2.2, Table 4: The GL-BP-GVF2 model has a much higher DIC (−4226.69) than GL-BP-GVF1 (−4265.64), yet the text does not discuss why adding the projected household covariate (z_i) worsens the model fit. A brief comment on this would be helpful.
  6. Equation (15): The absolute scaled discrepancy uses 'dz_i^T η' but the notation 'dz' is not defined. It appears to refer to the covariate vector in the GVF, but this should be stated explicitly.
  7. The manuscript states the date 'July 10, 2026' which appears to be a future date; this may be a typo.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; derivation is self-contained with minor self-citation to Hamura et al. (2024) for supporting lemmas

full rationale

The paper's core theoretical results (Corollary 2.1, Proposition 2.1, Proposition 2.2, Theorem 2.1, Theorem 3.1) are derived from first principles using the model structure (Eqs. 1, 4, 5) and standard asymptotic/analytic techniques (DCT, MCT, Gamma function bounds). The BP prior hyperparameters (a=2, b=1/2) are selected based on the theoretical properties in Corollary 2.1, not by fitting to data. The prior α~Ga(1,1) is a standard weakly-informative choice. The simulation design (Section 4) generates σ²_i in Exp-GVF form, which does favor the GL-BP model's shrinkage structure—but this is a question of simulation fairness (a correctness/robustness concern), not circularity: the simulation outcomes are not used to define or calibrate the model, and the theoretical results do not depend on the simulation settings. The self-citation to Hamura et al. (2024) is used for the proof technique of Theorem S1 (adapted in Corollary 2.1) and for the MH algorithm of Miller (2019), but the paper independently re-derives the results for the SAE context (different shrinkage target: Exp-GVF vs. grand mean) and notes substantive differences (Section 2.1). The posterior propriety result (Theorem 3.1) is proved from scratch. No 'prediction' reduces to a fitted input by construction, and no uniqueness theorem is invoked to forbid alternatives. The skeptic's concern about simulation design inflating ASD improvements is a validity concern, not a circularity concern—the simulation results do not feed back into the model definition or theoretical claims.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

No new entities are invented. The model uses established components: the Fay-Herriot model, Gamma sampling distribution for DVEs, Inverse-Gamma prior for variances, Beta Prime prior for local parameters, and GVF log-linear regression. The contribution is the specific combination and theoretical analysis within the SAE context.

free parameters (7)
  • a (BP shape 1) = 2
    Chosen based on Corollary 2.1 to satisfy properties P1-P3; not data-fitted but hand-selected from theory.
  • b (BP shape 2) = 0.5
    Chosen to produce spike at Exp-GVF (P3); same theoretical basis.
  • a_α (Gamma prior shape for α) = 1
    Standard weakly-informative choice for global parameter.
  • b_α (Gamma prior rate for α) = 1
    Standard weakly-informative choice for global parameter.
  • η (GVF regression coefficients)
    Estimated from data via MCMC; not free in the model-specification sense but data-driven.
  • β (FH regression coefficients)
    Estimated from data; standard FH model parameter.
  • σ²_u (random effects variance)
    Estimated from data; standard FH model parameter.
assumptions (5)
  • domain assumption D_i | σ²_i ~ Ga(ν_i/2, ν_i/(2σ²_i)) with ν_i = n_i - 1
    Section 1, Eq. 4. Justified under simple random sampling; approximated via Delta method for complex surveys in Section 5.2.1.
  • domain assumption The GVF log-linear model log(D_i) = z_i^T η + ε_i adequately captures the relationship between sampling variances and covariates
    Section 1, Eq. 3. Standard assumption in SAE practice following Wolter (2007).
  • domain assumption Covariates z_ik have the same signs for all areas i, for each k
    Theorem 3.1. Required for posterior propriety; satisfied when using log(n_i) as covariate.
  • standard math m - p - 2 > 0 (number of areas exceeds number of covariates plus 2)
    Theorem 3.1. Standard condition for posterior propriety in FH-type models.
  • domain assumption n_i > 1 for all areas
    Theorem 3.1. Ensures degrees of freedom are positive; excludes areas with single observations.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On improving the estimates of the sampling variances via Global-Local priors in Small Area Estimation." pith.science (2026). https://pith.science/paper/JB5G4GRG

@misc{pith2026260708720,
  author       = {Pith},
  title        = {Pith review of: On improving the estimates of the sampling variances via Global-Local priors in Small Area Estimation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JB5G4GRG}},
  note         = {Machine review of arXiv:2607.08720}
}
read the original abstract

The Fay-Herriot (FH) model is widely used in official statistics to produce reliable estimates for domains with small sample sizes. In the classical FH model, the sampling variances are treated as known, even though they are typically estimated from the data. In practice, these variance estimates can be highly variable. To address this issue, practitioners often use Generalized Variance Functions (GVFs) to borrow strength across areas and stabilize estimation. In this work, we propose a new Bayesian model to improve the posterior estimation of the sampling variances in Small Area Estimation (SAE). Our proposed model incorporates Global-Local (GL) priors to improve the level of shrinkage toward the posterior estimates obtained with the GVF function. We study the theoretical properties of the proposed model and develop adaptive Markov chain Monte Carlo (MCMC) algorithms to address computational challenges arising from conditional distributions involving Gamma functions. The performance of the proposed Bayesian model is investigated through simulation studies and compared with competing approaches. Finally, we implement our proposal in two real applications by estimating the Corn production from the U.S. Department of Agriculture and the Prevalence of the Educational Attainment Index at the municipality levels in Colombia.

Figures

Figures reproduced from arXiv: 2607.08720 by the authors.

Figure 1
Figure 1. Marginal prior distributions of σ 2 i under the Beta Prime (BP) and Gamma local priors. The right figure focuses on the tail region. In [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Marginal distributions of Di obtained under the proposed models using the Beta Prime (BP) and Gamma local priors and the model proposed by Maples et al. (2009). The plots on the bottom focus on the tail region on a logarithmic scale. In [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Direct Variance Estimates (DVEs) versus posterior estimates of [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Posterior estimates of p σ 2 i (left) and θi (right) in the Corn data. The red dots on the left and right plots are the squared roots of the Direct Variance Estimates (DVEs) and the direct survey estimates (yi), respectively. the EAI are available from the UNESCO Insti…
Figure 5
Figure 5. Figure 5: Direct estimates of the Prevalence of the Educational Attainment Index–At [PITH_FULL_IMAGE:figures/full_fig_p019_5.png]
Figure 6
Figure 6. Figure 6: Log Direct Variance Estimates (log DVEs) versus covariates considered in the [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: Comparison of square roots of Direct Variance Estimates (DVEs) and GVF [PITH_FULL_IMAGE:figures/full_fig_p021_7.png]
Figure 8
Figure 8. Figure 8: Direct Variance Estimates (DVEs) versus posterior estimates of [PITH_FULL_IMAGE:figures/full_fig_p023_8.png]
Figure 9
Figure 9. Figure 9: Direct estimates of the PEAI–AHS (yi) versus posterior estimates of θi in the PEAI–AHS application (left). The solid black line represents yi . Ratios of Coefficients of Variations (CVs) obtained with the GL-BP-GVF1 model and the direct estimator (right). of the Preval…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

  1. [1]

    E., Harter, R

    Battese, G. E., Harter, R. M., and Fuller, W. A. (1988). An E rror- C omponents M odel for P rediction of C ounty C rop A reas U sing S urvey and S atellite D ata. Journal of the American Statistical Association , 83(401):28--36

    work page 1988

  2. [2]

    and da Cunha, M

    Borchers, J. and da Cunha, M. S. (2025). School performance and inequality of opportunities in L atin A merica. Studies in Educational Evaluation , 86

    work page 2025

  3. [3]

    K., Kang, J., and Mukherjee, B

    Boss, J., Datta, J., Wang, X., Park, S. K., Kang, J., and Mukherjee, B. (2024). Group Inverse-Gamma Gamma Shrinkage for Sparse Linear Models with Block-Correlated Regressors . Bayesian Analysis , 19(3):785 -- 814

    work page 2024

  4. [4]

    Brown, P. J. and Griffin, J. E. (2010). Inference with normal-gamma prior distributions in regression problems . Bayesian Analysis , 5(1):171 -- 188

    work page 2010

  5. [5]

    M., Polson, N

    Carvalho, C. M., Polson, N. G., and Scott, J. G. (2010). The horseshoe estimator for sparse signals. Biometrika , 97(2):465--480

    work page 2010

  6. [6]

    Cox, C., Attewell, P., and Newman, K. (2010). E ducational I nequality in L atin A merica. Growing gaps. Educational inequality around the world , pages 33--58

    work page 2010

  7. [7]

    C., Maiti, T., Ren, H., and Sinha, S

    Dass, S. C., Maiti, T., Ren, H., and Sinha, S. (2012). Confidence interval estimation of small area parameters shrinking both means and variances. Survey Methodology , 38(2):173--187

    work page 2012

  8. [8]

    Encuesta nacional de demograf \'i a y salud 2015 (ends 2015)

    DHS (2015). Encuesta nacional de demograf \'i a y salud 2015 (ends 2015)

    work page 2015

Show all 36 references
  1. [9]

    Measuring rural poverty with a multidimensional approach: T he R ural M ultidimensional P overty I ndex

    FAO-OPHI (2022). Measuring rural poverty with a multidimensional approach: T he R ural M ultidimensional P overty I ndex. Technical Report Statistical Development Series No. 19, Food and Agriculture Organization of the United Nations (FAO), Rome

    2022

  2. [10]

    and Herriot, R

    Fay, R. and Herriot, R. (1979). E stimates of I ncome for S mall P laces: A n A pplication of J ames- S tein P rocedures to C ensus D ata. Journal of the American Statistical Association , 74(366):269--277

    1979

  3. [11]

    A n E ssay on the L ogical F oundations of S urvey S ampling, P art O ne

    H \'a jek, J. (1971). Comment on “ A n E ssay on the L ogical F oundations of S urvey S ampling, P art O ne”. In Godambe, V. P. and Sprott, D. A., editors, The Foundations of Survey Sampling , page 236. Holt, Rinehart, and Winston, New York

    1971

  4. [12]

    Hamura, Y., Onizuka, T., Hashimoto, S., and Sugasawa, S. (2024). S parse B ayesian I nference on G amma- D istributed O bservations U sing S hape- S cale I nverse- G amma M ixtures. Bayesian Analysis , 19(1):77--97

    2024

  5. [13]

    and Jedrzejczak, A

    Kubacki, J. and Jedrzejczak, A. (2012). The C omparison of G eneralized V ariance F unction with O ther M ethods of P recision E stimation for P olish H ousehold B udget S urvey. Studia Ekonomiczne , 120:58--69

    2012

  6. [14]

    Maiti, T., Ren, H., and Sinha, S. (2014). Prediction E rror of S mall A rea P redictors S hrinking B oth M eans and V ariances. Scandinavian Journal of Statistics , 41(3):775--790

    2014

  7. [15]

    Maples, J., Bell, W., and Huang, E. T. (2009). Small A rea V ariance M odeling with A pplication to C ounty P overty E stimates from the A merican C ommunity S urvey. Proceedings of the American Statistical Association Section on Survey Research Methods , pages 5056--5067

    2009

  8. [16]

    McIllece, J. J. (2018). On G eneralized V ariance F unctions for S ample M eans and M edians. JSM2018-survey research methods section , pages 584--594

    2018

  9. [17]

    Miller, J. W. (2019). Fast and A ccurate A pproximation of the F ull C onditional for G amma S hape P arameters. Journal of Computational and Graphical Statistics , 28(2):476--480

    2019

  10. [18]

    D., Pérez, A., and Hobza, T

    Morales, D., Esteban, M. D., Pérez, A., and Hobza, T. (2021). A Course on Small Area Estimation and Mixed Models: Methods, Theory and Applications in R . Statistics for Social and Behavioral Sciences, Springer

    2021

  11. [19]

    and Casella, G

    Park, T. and Casella, G. (2008). The B ayesian L asso. Journal of the american statistical association , 103(482):681--686

    2008

  12. [20]

    R., and Ram \'i rez, I

    P \'e rez, M.-E., Pericchi, L. R., and Ram \'i rez, I. C. (2017). The S caled B eta2 D istribution as a R obust P rior for S cales. Bayesian Analysis , 12(3):615--637

    2017

  13. [21]

    and Lleras, C

    Rangel, C. and Lleras, C. (2010). Educational inequality in C olombia: family background, school quality and student achievement in C artagena. International Studies in Sociology of Education , 20(4):291--317

    2010

  14. [22]

    Rivest, L. P. and Vandal, N. (2002). Mean squared error estimation for small areas when the small area variances are estimated. Proceedings of the International Conference on Recent Advances in Survey Sampling , pages 10--13

    2002

  15. [23]

    F., Moura, F

    Souza, D. F., Moura, F. A. S., and Migon, H. S. (2009). Small area population prediction via hierarchical models. Survey Methodology , 35:203 –214

    2009

  16. [24]

    J., Best, N

    Spiegelhalter, D. J., Best, N. G., Carlin, B. P., and Van Der Linde, A. (2002). Bayesian M easures of M odel C omplexity and F it. Journal of the Royal Statistical Society Series B: Statistical Methodology , 64(4):583--639

    2002

  17. [25]

    Sugasawa, S., Tamae, H., and Kubokawa, T. (2017). Bayesian E stimators for S mall A rea M odels S hrinking B oth M eans and V ariances. Scandinavian Journal of Statistics , 44(1):150--167

    2017

  18. [26]

    S., and Sedransk, J

    Tang, X., Ghosh, M., Ha, N. S., and Sedransk, J. (2018). Modeling R andom E ffects U sing G lobal- L ocal S hrinkage P riors in S mall A rea E stimation. Journal of the American Statistical Association , 113(524):1476--1489

    2018

  19. [27]

    The S ustainable D evelopment G oals E xtended R eport 2025: I nputs and information provided as of 30 april 2025

    UN (2025). The S ustainable D evelopment G oals E xtended R eport 2025: I nputs and information provided as of 30 april 2025. https://unstats.un.org/sdgs/report/2025/extended-report/. Accessed: 2025‑12‑07

    2025

  20. [28]

    World Education Indicators 2005 Education Trends in Perspective: Education Trends in Perspective

    UNESCO (2005). World Education Indicators 2005 Education Trends in Perspective: Education Trends in Perspective . OECD publishing

    2005

  21. [29]

    Transforming our world: the 2030 A genda for S ustainable D evelopment

    United Nations (2015). Transforming our world: the 2030 A genda for S ustainable D evelopment. R esolution adopted by the G eneral A ssembly on 25 S eptember 2015. https://sustainabledevelopment.un.org/post2015/transformingourworld. Accessed: 2025‑12‑07

    2015

  22. [30]

    and Fuller, W

    Wang, J. and Fuller, W. A. (2003). The M ean S quared E rror of S mall A rea P redictors C onstructed with E stimated A rea V ariances. Journal of the American Statistical Association , 98(1):716--723

    2003

  23. [31]

    Wolter, K. M. (2007). Introduction to Variance Estimation: Statistics for Social Science and Behavorial Sciences . Springer New York, New York, NY

    2007

  24. [32]

    Education S tatistics ( E d S tats): UIS D ata

    World-Bank (2024). Education S tatistics ( E d S tats): UIS D ata. https://databank.worldbank.org/source/education-statistics-(uis). World Bank EdStats database containing education indicators sourced from the UNESCO Institute for Statistics (UIS)

    2024

  25. [33]

    You, Y. (2021). Small area estimation using F ay- H erriot area level model with sampling variance smoothing and modeling. Survey Methodology , 47(2):361--370

    2021

  26. [34]

    and Chapman, B

    You, Y. and Chapman, B. (2006). Small A rea E stimation U sing A rea L evel M odels and E stimated S ampling V ariances. Survey Methodology , 32:97--103

    2006

  27. [35]

    and Hidiroglou, M

    You, Y. and Hidiroglou, M. A. (2023). Application of S ampling V ariance S moothing M ethods for S mall A rea P roportion E stimation. Journal of Official Statistics , 39(4):571--590

    2023

  28. [36]

    Zhang, G., Cheng, Y., and Lu, Y. (2019). Generalised variance functions for longitudinal survey data. Statistical Theory and Related Fields , 3(2):150--157

    2019

Pith tools

Reviewed July 10, 2026 · model on record in the stance chip above.