Prior distributions for variance parameters in hierarchical models (comment on article by Browne and Draper)
Open accessThe DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index; 1 of the 23 sources its claims cite were checked for support.
Abstract
Various noninformative prior distributions have been suggested for scale parameters in hierarchical models. We construct a new folded-noncentral-$t$ family of conditionally conjugate priors for hierarchical standard deviation parameters, and then consider noninformative and weakly informative priors in this family. We use an example to illustrate serious problems with the inverse-gamma family of "noninformative" prior distributions. We suggest instead to use a uniform prior on the hierarchical standard deviation, using the half-$t$ family when the number of groups is small and in other settings where a weakly informative prior is desired. We also illustrate the use of the half-$t$ family for hierarchical modeling of multiple variance parameters such as arise in the analysis of variance.
Overview
CoChat Summary
TL;DR
Gelman argues that inverse-gamma “noninformative” priors for hierarchical variance parameters can badly distort posterior inference, especially when the group-level variance is near zero or the number of groups is small. He recommends a uniform prior on the hierarchical standard deviation when groups are not few, and half-t/half-Cauchy weakly informative priors otherwise, developed within a folded-noncentral-t conditionally conjugate family.
Key findings
- The inverse-gamma(ε, ε) family is not reliably noninformative: as ε → 0 it yields an improper posterior, and in the 8-schools example inverse-gamma(1, 1) and inverse-gamma(0.001, 0.001) priors on σ²_α constrained posterior inferences and increased shrinkage relative to a uniform prior on σ_α.
- A uniform prior on σ_α performed well for the 8-schools data (J = 8), but with only J = 3 schools it produced an extremely long right tail and undershrinkage; a half-Cauchy prior with scale A = 25 removed much of that unrealistic upper tail.
- For multiple variance components, Gelman recommends a hierarchical half-Cauchy model with scale estimated from data; in the 5 × 5 × 2 split-plot latin square Anova with nine variance components, the Cauchy scale parameter A was estimated at 1.8 with a 95% posterior interval of [0.5, 5.1].
- The folded-noncentral-t family, obtained via a redundant multiplicative parameterization, provides conditionally conjugate priors for σ_α and includes the improper uniform density (ν = −1) and proper half-Cauchy (ν = 1) as special cases.
Methods
- Conceptual/comment paper using a simple two-level normal hierarchical model: y_ij ∼ N(μ + α_j, σ²_y), α_j ∼ N(0, σ²_α), focusing on the prior for σ_α.
- Illustrates with the 8-schools example (J = 8, 6000 Bugs simulation draws), a 3-schools subset, and a 5 × 5 × 2 split-plot latin square Anova with nine variance components.
- Uses a parameter-expanded form y_ij ∼ N(μ + ξη_j, σ²_y), η_j ∼ N(0, σ²_η), with σ_α = |ξ|σ_η, to construct and compare uniform, inverse-gamma, half-Cauchy, and hierarchical half-Cauchy priors.
Limitations
- The paper states that noninformative and weakly informative priors are inherently provisional: after fitting, one should inspect the posterior and revise the prior if it does not make sense.
- The paper notes that the uniform prior on σ_α has a positive miscalibration and yields an improper posterior for J = 1 or 2; inverse-gamma inferences are sensitive to ε; and half-Cauchy scale choices are weakly informative rather than fully objective.
- Evident from the text: the recommendations rest on illustrative examples and conceptual argument rather than a broad simulation study or systematic empirical comparison, and extensions to nonlinear, multivariate, and more complex hierarchical models are left for further work.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 14 Not checked · 56 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Similarly, there are no always-nonnegative classical unbiased estimators of σ α or σ 2 α in the hierarchical model.”
- Also stated in: § 7.1 Prior distributions for variance parameters · plain
“Similarly, there are no alwaysnonnegative classical unbiased estimators of σ α or σ 2 α in the hierarchical model.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 2.4 Calibration
What the sources say
- Could not verifycites Bickel and Blackwell (1967) matched to “A Note on Bayes Estimates” (1967) sourceExperimental — accuracy not yet measured
The introduction (p. 1) and Theorem 1.3 (p. 3) concern when an unbiased Bayes estimate must have zero Bayes risk. Searches for hierarchical models, σₐ, σ²ₐ, and always-nonnegative classical unbiased estimators found no passage addressing the claimed nonexistence result. (LLM, conf=0.77)
- Not checked — no full text of the sourcecites Meng and Zaslavsky (2002) matched to “Single observation unbiased priors” (2002) source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 7.1 Prior distributions for variance parameters · plain
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“If J is low, the uniform prior density tends to lead to high estimates of σ α , as discussed in Section 5.2.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 7.1 Prior distributions for variance parameters
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In contrast, the hierarchical half-Cauchy model performs a great deal of shrinkage, especially of the high ranges of the intervals.”
- Also stated in: § Bugs model: a half-Cauchy prior distribution on sigma.theta is induced # using a normal prior on xi and an inverse-gamma on tau.eta model { for (j in 1:J){ # J = the number of schools y[j] ~dnorm (theta[j], tau.y[j]) # data model: the likelihood theta[j] <-mu.theta+ xi*eta[j] tau.y[j] <-pow(sigma.y[j],-2) } xi ~dnorm (0, tau.xi) tau.xi <-pow(prior.scale,-2) for (j in 1:J){ eta[j] ~dnorm (0, tau.eta) # hierarchical model for theta } tau.eta ~dgamma (.5, .5)# chi^2 with 1 d.f.sigma.theta<-abs(xi)/sqrt(tau.eta)# cauchy = normal/sqrt(chi^2) mu.theta ~dnorm (0.0, 1.0E-6) # noninformative prior on mu } · plain
“The half-Cauchy model gives much sharper inferences, using the partial pooling that comes with fitting a hierarchical model.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 6.2 Superpopulation and finite-population standard deviations
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Bugs model: a half-Cauchy prior distribution on sigma.theta is induced # using a normal prior on xi and an inverse-gamma on tau.eta model { for (j in 1:J){ # J = the number of schools y[j] ~dnorm (theta[j], tau.y[j]) # data model: the likelihood theta[j] <-mu.theta+ xi*eta[j] tau.y[j] <-pow(sigma.y[j],-2) } xi ~dnorm (0, tau.xi) tau.xi <-pow(prior.scale,-2) for (j in 1:J){ eta[j] ~dnorm (0, tau.eta) # hierarchical model for theta } tau.eta ~dgamma (.5, .5)# chi^2 with 1 d.f.sigma.theta<-abs(xi)/sqrt(tau.eta)# cauchy = normal/sqrt(chi^2) mu.theta ~dnorm (0.0, 1.0E-6) # noninformative prior on mu } · plain
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This prior distribution is even more sharply peaked near zero and further distorts posterior inferences, with the problem arising because the marginal likelihood for σ α remains high near zero.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 5.1 Noninformative prior distributions for the 8-schools problem
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The uniform prior distribution seems fine for the 8-school analysis”
- Also stated in: § 5.2 Weakly informative prior distribution for the 3-schools problem · plain
“the default uniform prior gave reasonable results”
- Also stated in: § 7.1 Prior distributions for variance parameters · hedged
“Figure 1 illustrates the generally robust properties of the uniform prior density on σ α .”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 5.2 Weakly informative prior distribution for the 3-schools problem
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 5.2 Weakly informative prior distribution for the 3-schools problem · plain
26 more claims
Create a free account to see all 71 claims and every source verdict.
- It may state two separate claims as one. 34
- Probably not a substantial claim. 12
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Because the uniform distribution tends to unduly favor presence of heterogeneity in discrimination and calibration estimates across studies,58,59”
From A framework for meta-analysis of prediction model studies with binary and time-to-event outcomes (2019) · cites this paper as [59]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 7.1 Prior distributions for variance parameters
“we see conditional conjugacy in the wider class of half-t distributions on σ α , which include the uniform and half-Cauchy densities on σ α (as well as inverse-gamma on σ 2 α ) as special cases.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 3.2 Folded-noncentral-t prior distribution for σ α
“In the limit that the denominator is specified exactly, we have a folded normal distribution; conversely, specifying the numerator exactly yields the square-root-inverse-χ 2 distribution for σ α , as in Section 3.1.”
Citations
Cites
Gelman, A., Carlin, J. B., Stern, H. S., & Rubin, D. B. (2003) · doi:10.1201/9780429258480 · Cited by 27,276
Self-citationKass, R. E., & Raftery, A. E. (1995) · Journal of the American Statistical Association · doi:10.1080/01621459.1995.10476572 · Cited by 12,558
Sampling-based approaches to calculating marginal densities
Gelfand, A. E., & Smith, A. F. M. (1990) · Journal of the American Statistical Association · doi:10.1080/01621459.1990.10476213 · Cited by 6,606 · cited 1×
“Typically, enough data will be available to estimate µ and σy that one can use any reasonable noninformative prior distribution—for example, p(µ, σy) ∝ 1 or p(µ, log σy) ∝ 1.”
Introducing Multilevel Modeling
Kreft, I., & de Leeuw, J. (1998) · doi:10.4135/9781849209366 · Cited by 2,997
POSTERIOR PREDICTIVE ASSESSMENT OF MODEL FITNESS VIA REALIZED DISCREPANCIES
Gelman, A., Meng, X., & Stern, H. S. (1996) · Cited by 2,638
Self-citationBayes and Empirical Bayes Methods for Data Analysis
(2001) · Technometrics · doi:10.1198/tech.2001.s608 · Cited by 2,098
INADMISSIBILITY OF THE USUAL ESTIMATOR FOR THE MEAN OF A MULTIVARIATE NORMAL DISTRIBUTION
Stein, C. (1956) · Contribution to the Theory of Statistics · doi:10.1525/9780520313880-018 · Cited by 1,841
Bayesian Approaches to Clinical Trials and Health‐Care Evaluation
Spiegelhalter, D. J., Abrams, K. R., & Myles, J. P. (2003) · Wiley eBooks · doi:10.1002/0470092602 · Cited by 1,673
Parametric empirical Bayes inference: theory and applications
Morris, C. N. (1983) · Journal of the American Statistical Association · doi:10.1080/01621459.1983.10477920 · Cited by 1,448 · cited 1×
“On the theoretical side, hierarchical models allow a more “objective” approach to inference by estimating the parameters of prior distributions from data rather than requiring them to be specified using subjective information (see James and Stein, 1960, Efron and Morris, 1975, and Morris, 1983).”
The Selection of Prior Distributions by Formal Rules
Kass, R. E., & Wasserman, L. (1996) · Journal of the American Statistical Association · doi:10.1080/01621459.1996.10477003 · Cited by 1,308
Research works printed but not linked: 5
- Box, G. E. P. and Tiao, G. C. (1973). Bayesian Inference in Statistical Analysis. Reading, Mass.: Addison-Wesley.
- James, W. and Stein, C. (1960). "Estimation with quadratic loss." In Neyman, J. (ed.), Proceedings of the Fourth Berkeley Symposium, volume 1, 361–380. Berkeley: University of California Press.
- Johnson, N. L. and Kotz, S. (1972). Distributions in Statistics. New York: Wiley. 4 vols.
- Liu, C. (2001). "Bayesian analysis of multivariate probit models. Discussion of “The art of data augmentation” by D. A. van Dyk and X. L. Meng." Journal of Computational and Graphical Statistics, 10: 75–81.
- Raudenbush, S. W. and Bryk, A. S. (2002). Hierarchical Linear Models. Thousand Oaks, Calif.: Sage., second edition.
Cited by
Items 2,101–2,200 of 4,089
Charpentier, E., Looten, V., Fahlgren, B., Barna, A., & Guillevin, L. (2016) · BMC Medical Research Methodology · doi:10.1186/s12874-016-0107-5 · Cited by 8
Bayesian first order auto-regressive latent variable models for multiple binary sequences
Giardina, F., Guglielmi, A., Quintana, F. A., et al. (2011) · Statistical Modelling · doi:10.1177/1471082x1001100601 · Cited by 8
“In contrast, estimation of regression coefficients and predictions are fairly sensitive to different choices of ã and b̃; this was still true when changing the prior from Gamma to a uniform on 1/ √ τ , as suggested in Gelman (2006).”
MethodTorabi, M. (2016) · Biometrical Journal · doi:10.1002/bimj.201500248 · Cited by 8
“…deviations can be uniformly C© 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.biometrical-journal.com distributed, because of the robust properties of this prior (Gelman, 2006); corresponding priors to mixing parameters (πi j1, . . . , πi jL) in (4), which are through υ jl and ζi jl ,…”
BackgroundImproved component predictions of batting and pitching measures
Albert, J. (2016) · Journal of Quantitative Analysis in Sports · doi:10.1515/jqas-2015-0063 · Cited by 8
Multiscale measurement error models for aggregated small area health data
Aregay, M., Lawson, A. B., Faes, C., Kirby, R. S., Carroll, R. M., & Watjou, K. (2016) · Statistical Methods in Medical Research · doi:10.1177/0962280216661094 · Cited by 8
A Gaussian random field model for similarity-based smoothing in Bayesian disease mapping
Baptista, H., Mendes, J. M., MacNab, Y. C., Xavier, M., & Caldas‐de‐Almeida, J. M. (2016) · Statistical Methods in Medical Research · doi:10.1177/0962280216660407 · Cited by 9
Early‐life density‐dependence effects on growth and survival in subantarctic fur seals
Pacoureau, N., Authier, M., Delord, K., Guinet, C., & Barbraud, C. (2017) · Population Ecology · doi:10.1007/s10144-017-0573-6 · Cited by 8
“In absence of independent information to specify informative priors, weakly informative priors were used: Student or normal distributions (Gelman 2006).”
MethodIijima, H. (2018) · Mammal Study · doi:10.3106/ms2017-0087 · Cited by 8
“The prior distribution of s1 was a vague uniform distribution of U(0, 100) (Gelman 2006).”
BackgroundHu, J. M. Y., Zhuang, L. H., Bernardo, B. A., & McCandless, L. C. (2018) · Current Epidemiology Reports · doi:10.1007/s40471-018-0156-x · Cited by 10
Bootstrap Prediction Intervals for Small Area Means from Unit-Level Nonlinear Models
Erciulescu, A. L., & Fuller, W. A. (2018) · Journal of Survey Statistics and Methodology · doi:10.1093/jssam/smy014 · Cited by 8
Saying Sayonara to the Farm: Hierarchical Bayesian Modeling of Farm Exits in Japan
Ramsey, A. F., Ghosh, S. K., & Sonoda, T. (2018) · Journal of Agricultural Economics · doi:10.1111/1477-9552.12290 · Cited by 8
Pavement degradation: a city-scale model for San Francisco, USA
Zhao, B., Silva, E. A., & Soga, K. (2018) · Proceedings of the Institution of Civil Engineers - Smart Infrastructure and Construction · doi:10.1680/jsmic.18.00001 · Cited by 8
Direct and indirect genetic effects on reproductive investment in a grasshopper
Chakrabarty, A., van Kronenberg, P., Toliopoulos, N., et al. (2019) · Journal of Evolutionary Biology · doi:10.1111/jeb.13417 · Cited by 8
“For models fitted in MCMCglmm, we used dispersed default priors for fixed effects and parameter expanded (half- Cauchy) priors for the random effects (Gelman, 2006).”
MethodGershunskaya, J., & Savitsky, T. D. (2019) · Journal of Survey Statistics and Methodology · doi:10.1093/jssam/smz004 · Cited by 9
Multivariate air pollution prediction modeling with partial missingness
Boaz, R. M., Lawson, A. T. D.-D., & Pearce, J. L. (2019) · Environmetrics · doi:10.1002/env.2592 · Cited by 8
“The inverse gamma priors are conditionally conjugate and weakly informative (Gelman, 2006).”
BackgroundUsing History Matching for Prior Choice
Wang, X., Nott, D. J., Drovandi, C., Mengersen, K. L., & Evans, M. J. (2018) · Technometrics · doi:10.1080/00401706.2017.1421587 · Cited by 10
“…particular found use in the checking for prior-data conflicts when the summary statistic is a minimal sufficient statistic (Evans and Moshonov 2006) and for giving a precise formulation of the notion of a weakly informative prior (as in Evans and Jang 2011, inspired by earlier work of Gelman 2006).”
BackgroundMethodInfluentialBreaking the link? How European integration shapes social policy demand and supply
Tober, T., & Busemeyer, M. R. (2020) · Journal of European Public Policy · doi:10.1080/13501763.2020.1824010 · Cited by 10
“Given that the number of groups is relatively small, we assign weakly informative half-t priors, t(4, 0, 1), on the variance components (Gelman, 2006).”
MethodA Two-level Moderated Latent Variable Model with Single Level Data
Liu, H., Yuan, K., & Liu, F. (2019) · Multivariate Behavioral Research · doi:10.1080/00273171.2019.1689350 · Cited by 9
Bayesian Reliability Analysis of Marshall and Olkin Model
AbuJarad, M. H. A., Khan, A. A., Khaleel, M. A., et al. (2019) · Annals of Data Science · doi:10.1007/s40745-019-00234-3 · Cited by 8
A group-specific prior distribution for effect-size heterogeneity in meta-analysis
Thompson, C. G., & Becker, B. J. (2020) · Behavior Research Methods · doi:10.3758/s13428-020-01382-8 · Cited by 8
“In addition to the FNT, other choices of priors for τjwere a) the half-Cauchy as demonstrated in Gelman (2006), b) the half-normal with large variance, c) the inverse-gamma with small and identical shape and scale parameters, and d) the uniform with a wide range.”
BackgroundMethodInfluentialA Bayesian multivariate meta‐analysis of prevalence data
Siegel, L. K., Rudser, K. D., Sutcliffe, S., Markland, A. D., Brubaker, L. T., Gahagan, S., Stapleton, A. E., & Chu, H. (2020) · Statistics in Medicine · doi:10.1002/sim.8593 · Cited by 9
Millar, J. J., Toh, K. B., & do Valle, D. R. (2020) · BMC Medicine · doi:10.1186/s12916-020-01609-7 · Cited by 8
“Ghana Demographic Health Survey [50] September 2014–December 2014 2713 10”
BackgroundInfluentialWalsworth, T. E., & Schindler, D. E. (2015) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/cjfas-2014-0554 · Cited by 8
“Broad, bounded prior distributions were used for all hyper-parameters because we had little information to otherwise inform them (mean values for the river and among year standard deviation; Table 1; Gelman 2006).”
MethodSequential Hierarchical Regression Imputation
Yucel, R. M., Zhao, E., Nathaniel Schenker, A., & Raghunathan, T. E. (2017) · Journal of Survey Statistics and Methodology · doi:10.1093/jssam/smx004 · Cited by 8
Estimating heterogeneity variances to select a random effects model
Rukhin, A. L. (2019) · Journal of Statistical Planning and Inference · doi:10.1016/j.jspi.2018.12.003 · Cited by 8
Thompson, F. N., Oliver, M. D., & Abraham, E. R. (2009) · Cited by 8
Bayesian inference of Gaussian mixture models with noninformative priorsPreprint
Stoneking, C. J. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1405.4895 · Cited by 8
“Besides the Jeffreys prior mentioned thus far, uniform priors for variance parameters would also be possible (Gelman, 2006).”
BackgroundA Cross-Cohort Changepoint Model for Customer-Base AnalysisPreprint
Gopalakrishnan, A., Bradlow, E. T., & Fader, P. S. (2012) · SSRN Electronic Journal · doi:10.2139/ssrn.2119337 · Cited by 8
Williams, D. R., Liu, S., Martin, S. R., et al. (2019) · doi:10.31234/osf.io/4kfjp · Cited by 8
“This number of iterations provided a good quality of the parameter estimates in which the models converged with potential scale reduction factors R smaller than 1.1 (Gelman, 2006).”
MethodInfluentialBeast: A Model for the Assessment of System-Wide Risks and Macroprudential PoliciesPreprint
Budnik, K. B., Größ, J., Vagliano, G., et al. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4599639 · Cited by 8
A Bayesian hierarchical model for combining multiple data sources in population size estimation
Parsons, J., Niu, X. M., & Bao, L. (2022) · The Annals of Applied Statistics · doi:10.1214/21-aoas1556 · Cited by 7
Nakashima, D., Fujii, K., Tsubouchi, Y., Kubo, Y., Yorozuya, K., Noritake, K., Tomiyama, N., Tsujishita, S., & Iitsuka, T. (2024) · Healthcare · doi:10.3390/healthcare12020167 · Cited by 10
The role of art knowledge training on aesthetic judgements and executive functions
Bara, I., Cross, E. S., & Ramsey, R. (2025) · Royal Society Open Science · doi:10.1098/rsos.240175 · Cited by 12
“Also, by using weakly informative priors, we allow for the possibility of large effects, should they exist in the data [109–112].”
Ye, Y., Zheng, P., Wang, Q., Wong, S. C., & Xu, P. (2025) · Analytic Methods in Accident Research · doi:10.1016/j.amar.2025.100384 · Cited by 7
Dahlke, J., & Ebersberger, B. (2025) · Journal of Evolutionary Economics · doi:10.1007/s00191-025-00909-6 · Cited by 9
“The model assumes a normal prior distribution for the topic-document distribution while pooling variance at the topic level, producing a half-Cauchy distribution as recommended for Bayesian models of this type (Gelman 2006; Polson and Scott 2012).”
Use and reporting of historical control data for regulatory studies
(PPR), E. P. O. P. P. P. A. T. R., Čoja, T., Adriaanse, P., Choi, J. M., Finizio, A., Giraudo, M., Kühl, T., Metruccio, F., Paparella, M., Pieper, S., et al. (2025) · EFSA Journal · doi:10.2903/j.efsa.2025.9576 · Cited by 15
Tian, Z., Zhou, S., Yin, X., et al. (2025) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2025.111633 · Cited by 7
Analysis of SF-6D Health State Utility Scores: Is Beta Regression Appropriate?
Kharroubi, S. A. (2020) · Healthcare · doi:10.3390/healthcare8040525 · Cited by 10
Factors affecting antler growth period and casting date in red deer
Gómez, J. Á., Pérez-Barbería, F. J., García, A. J., Cappelli, J., Chonco, L., Ceacero, F., Pérez-Serrano, M., & Landete‐Castillejos, T. (2021) · Journal of Mammalogy · doi:10.1093/jmammal/gyab097 · Cited by 11
Estimating genetic variance contributed by a quantitative trait locus: A random model approach
Wang, S., Xie, F., & Xu, S. (2022) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1009923 · Cited by 11
Luan, H., & Ransome, Y. (2022) · Annals of the American Association of Geographers · doi:10.1080/24694452.2022.2080040 · Cited by 7
Kraljič, A., Matkovič, A., Purg, N., Demšar, J., & Repovš, G. (2022) · Frontiers in Neuroimaging · doi:10.3389/fnimg.2022.968363 · Cited by 10
“…distributions (µ = 0, σ = 10) for regression parameters and half-Cauchy prior distributions (µ = 0, λ = 2.5) for standard deviations, as recommended by Gelman (2006), for all but the correlation model, where we used a vague normal (µ = 0, σ = 100) prior for the means of the input vectors, a…”
MethodInfluentialCorrea-Álvarez, C. D., Salazar-Uribe, J. C., & Pericchi-Guerra, L. R. (2022) · Computational Statistics · doi:10.1007/s00180-022-01287-4 · Cited by 7
“It is important to be careful when using the inverse-gamma distribution because studies such as those of Berger (2006), Gelman (2006), Pérez et al. (2017), and Rojas and Ramírez (2019) have out-lined the drawbacks of using it as a prior distribution.”
MethodInfluentialSorel, M. H., Murdoch, A. R., Zabel, R. W., Jorgensen, J. C., Kamphaus, C. M., & Converse, S. J. (2023) · Ecosphere · doi:10.1002/ecs2.4366 · Cited by 7
Subject-specific Dirichlet-multinomial regression for multi-district microbiota data analysis
Pedone, M., Amedei, A., & Stingo, F. C. (2023) · The Annals of Applied Statistics · doi:10.1214/22-aoas1641 · Cited by 8
Blozis, S. A. (2023) · Behavior Research Methods · doi:10.3758/s13428-023-02128-y · Cited by 7
“This kind of problem is analogous to fitting three-level models for which the number of random subjects at the highest level is small (Gelman, 2006).”
BackgroundBet Hedging Is Not Sufficient to Explain Germination Patterns of a Winter Annual Plant
Siegmund, G., Moeller, D. A., Eckhart, V. M., & Geber, M. A. (2023) · The American Naturalist · doi:10.1086/726785 · Cited by 6
A two-stage approach for Bayesian joint models: reducing complexity while maintaining accuracy
Alvares, D., & Leiva-Yamaguchi, V. (2023) · Statistics and Computing · doi:10.1007/s11222-023-10281-9 · Cited by 9
Evaluating Economic Impacts of Automation Using Big Data Approaches
Ardakani, O. M., & Sáenz, M. (2023) · Journal of Data Science and Intelligent Systems · doi:10.47852/bonviewjdsis32021569 · Cited by 7
Bayesian Inference for Multivariate Spatial Models with INLA
Palmí‐Perales, F., Gómez‐Rubio, V., Bivand, R., Cameletti, M., & Rue, H. (2023) · The R Journal · doi:10.32614/rj-2023-068 · Cited by 13
“For instance, Gelman (2006) suggests to avoid inverse Gamma distributions on the precision and propose some alternatives.”
BackgroundChinembiri, T. S., Mutanga, O., & Dube, T. J. (2024) · GIScience & Remote Sensing · doi:10.1080/15481603.2024.2303868 · Cited by 9
“The technique is a type of a MCMC method that generates sequentially correlated draws from a sequence of probability distributions (Gelman 2006).”
Fraterrigo, J. M., Chen, W., Loyal, J., & Euskirchen, E. S. (2024) · Journal of Ecology · doi:10.1111/1365-2745.14278 · Cited by 8
“We assigned all coefficients noninformative priors to minimise the influence of the prior on the posterior distribution (Gelman, 2006) and estimated the models using a Markov chain Monte Carlo algorithm with four chains run for 3000 iterations after a burn-in of 1000 iterations.”
MethodA model-based hierarchical Bayesian approach to Sholl analysis
VonKaenel, E. D., Feidler, A. M., Lowery, R. L., Andersh, K. M., Love, T., Majewska, A. K., & McCall, M. N. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae156 · Cited by 9
A tutorial on Bayesian structural equation modelling: Principles and applications
Qi-jin, C., Su, K., Feng, Y., Zhang, L., Ding, R., & Pan, J. (2024) · International Journal of Psychology · doi:10.1002/ijop.13258 · Cited by 14
eDNAPlus: A Unifying Modeling Framework for DNA-based Biodiversity Monitoring
Diana, A., Matechou, E., Griffin, J. E., Yu, D. W., Luo, M., Tosa, M. I., Bush, A., & Griffiths, R. A. (2024) · Journal of the American Statistical Association · doi:10.1080/01621459.2024.2412362 · Cited by 23
Bayesian model inference: Exploring parsimonious models with MCMC and optimization approximations
Devashish, D., Farhin, S., Parvej, M., & khan, A. A. (2025) · Communication in Statistics- Theory and Methods · doi:10.1080/03610926.2025.2477826 · Cited by 8
“Gelman (2006) support the use of half-Cauchy prior for various parameters.”
Bayesian sensitivity models for missing covariates in the analysis of survival data
Hemming, K., & Hutton, J. L. (2010) · Journal of Evaluation in Clinical Practice · doi:10.1111/j.1365-2753.2010.01569.x · Cited by 7
Stamey, J. D., Natanegara, F., & Seaman, J. W. (2013) · Journal of Biopharmaceutical Statistics · doi:10.1080/10543406.2013.789885 · Cited by 7
Scheduling Viability Tests for Seeds in Long-Term Storage Based on a Bayesian Multi-Level Model
Trapp, A., Dixon, P. M., Widrlechner, M. P., & Kovach, D. A. (2012) · Journal of Agricultural Biological and Environmental Statistics · doi:10.1007/s13253-012-0085-y · Cited by 8
“We considered three prior distributions: (0.01,0.01), (0.001,0.001), and Uniform(0,10000) (Gelman 2006).”
InfluentialHuang, Z., Li, J., Nott, D. J., Feng, L., Ng, T.-P., & Wong, T. Y. (2014) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2014.928821 · Cited by 7
“[25] Gelman [26] suggests using the half-Cauchy distribution in place of inverse gamma priors for variance parameters when variance parameters are small and inferences become very sensitive to shape and scale parameters.”
MethodHelser, T. E., Kastelle, C. R., & Lai, H.-L. (2014) · Ecological Modelling · doi:10.1016/j.ecolmodel.2014.01.011 · Cited by 7
Outcome-Driven Cluster Analysis with Application to Microarray Data
Hsu, J. J., Finkelstein, D. M., & Schoenfeld, D. (2015) · PLoS ONE · doi:10.1371/journal.pone.0141874 · Cited by 7
Mixture Joint Models for Event Time and Longitudinal Data With Multiple Features
Huang, Y., Dagne, G. A., & Park, J. (2016) · Statistics in Biopharmaceutical Research · doi:10.1080/19466315.2016.1142891 · Cited by 7
“…distributions, we conducted a limited sensitivity analysis using the uniform prior distributionU (0, 100) for standard deviation scales σ1 and σ2 (Gelman 2006) instead of inverse gamma prior distribution (used in this article) for variance parameters σ 21 and σ 22 ; we refitted data based…”
MethodBayesian lower bounds for dense or sparse (Outlier) noise in the RMT framework
Ollier, V., Boyer, R., Korso, M. N. E., et al. (2016) · Proceedings of the IEEE Sensor Array and Multichannel Signal Processing Workshop · doi:10.1109/sam.2016.7569694 · Cited by 7
“In addition, if this hyper-parameter is Gamma distributed [17,18], then the marginalized joint pdf over the hyper-parameter is the Student’s t-distribution.”
BackgroundEffects of Prediction Error Bias on Model Calibration and Response Prediction of a 10-Story Building
Behmanesh, I., Yousefianmoghadam, S., Nozari, A., et al. (2016) · Conference proceedings of the Society for Experimental Mechanics · doi:10.1007/978-3-319-29754-5_28 · Cited by 7
“The error functions are modeled as Gaussian distributions [28–30]:”
BackgroundBayesian inference for sinh-normal/independent nonlinear regression models
Vilca, F., Azevedo, C., & Balakrishnan, N. (2016) · Journal of Applied Statistics · doi:10.1080/02664763.2016.1238058 · Cited by 7
“If no prior information is available in the form of historical data or from past experiments, we may assign conditional conjugate priors to the parameters, in the sense of Gelman [16].”
MethodJoint bent-cable Tobit models for longitudinal and time-to-event data
Dagne, G. A. (2017) · Journal of Biopharmaceutical Statistics · doi:10.1080/10543406.2017.1321006 · Cited by 7
“We conducted a sensitivity analysis using a uniform prior distribution (unif ð0; 100Þ) for σe (Gelman, 2006) instead of inverse gamma prior distribution (IGð0:01; 0:01Þ) for σ2e , a gamma distribution (gammað3:5; 3:5ÞIð2; Þ) for the degrees of freedom ν instead of an exponential prior distribution…”
MethodInclusion of biological knowledge in a Bayesian shrinkage model for joint estimation of SNP effects
Pereira, M., Thompson, J. R., Weichenberger, C. X., Thomas, D. C., & Minelli, C. (2017) · Genetic Epidemiology · doi:10.1002/gepi.22038 · Cited by 7
“We consider a model where the SNP effects follow a normal distribution and the variance is modeled by a shrinkage hyperprior (Gelman, 2006).”
InfluentialSparsity-based estimation bounds with corrupted measurements
Boyer, R., & Larzabal, P. (2017) · Signal Processing · doi:10.1016/j.sigpro.2017.08.004 · Cited by 7
Blomqvist, B. R. H., Mann, R. P., & Sumpter, D. J. T. (2018) · PLoS ONE · doi:10.1371/journal.pone.0196355 · Cited by 6
Nieto, R., & Casabianca, J. M. (2019) · Journal of Educational Measurement · doi:10.1111/jedm.12225 · Cited by 8
PC priors for residual correlation parameters in one-factor mixed models
Ventrucci, M., Cocchi, D., Burgazzi, G., & Laini, A. (2019) · Statistical Methods & Applications · doi:10.1007/s10260-019-00501-w · Cited by 12
“The literature on priors for variance parameters is vast, see for instance (Gelman 2006), and it is outside the scope of this work to give a comprehensive review of it.”
BackgroundChitakasempornkul, K., Rosa, G. J. M., Jager, A., & Bello, N. M. (2020) · Journal of Agricultural Biological and Environmental Statistics · doi:10.1007/s13253-020-00389-0 · Cited by 9
“This proxy is consistent with the prior √ σ 2 e j ∼ U (0, A), for any finite but appropriately large value of A such that the resulting distribution is vague, as recommended for variance components by Gelman (2006).”
BackgroundFlexible link functions in a joint hierarchical Gaussian process model
Su, W., Wang, X., & Szczesniak, R. D. (2020) · Biometrics · doi:10.1111/biom.13291 · Cited by 7
“For the scale parameters 𝜙𝑦, 𝜙𝑟, and 𝜙𝜓 , we use a truncated normal distribution with mean 0 and variance 52 as a weakly informative prior (Gelman, 2006; Stan Development Team, 2017); for the coefficient of the shared-parameter 𝑐, we use a uniform prior ranging from −1 to 1; for the variance…”
MethodConfidence Intervals for Coefficient of Variation of Inverse Gaussian Distribution
Chankham, W., Niwitpong, S., & Niwitpong, S. (2019) · International Conference on Vision, Image and Signal Processing (ICVISP) · doi:10.1145/3387168.3387254 · Cited by 7
Clustering and Prediction With Variable Dimension Covariates
Page, G. L., Quintana, F. A., & Müller, P. (2021) · Journal of Computational and Graphical Statistics · doi:10.1080/10618600.2021.1999824 · Cited by 7
Cabral, C. R. B., de Souza, N. L., & Leão, J. (2021) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2021.1969397 · Cited by 7
“[51], in hierarchical models, when using the inversegamma(ξ , ξ) noninformative prior distribution for variance parameters, the resulting inferences can be sensitive to ξ .”
MethodLiu, Y., Béliveau, A., Besche, H. C., Wu, A. D., Zhang, X., Stefan, M. I., Gutlerner, J. L., & Kim, C. (2021) · Frontiers in Education · doi:10.3389/feduc.2020.607260 · Cited by 6
“Third, it is recommended over the frequentist approach when the model is complex and the sample size is small (Browne and Draper, 2006; Gelman, 2006; Baldwin and Fellingham, 2013; Stegmueller, 2013).”
Adaptive Bayesian Spectral Analysis of High-Dimensional Nonstationary Time Series
Li, Z., Rosen, O., Ferrarelli, F., et al. (2021) · Journal of Computational and Graphical Statistics · doi:10.1080/10618600.2020.1868305 · Cited by 7
“Prior distributions on the error variances are placed by assuming that the σǫ,p, for p = 1, · · · , P , are independent Half-t(ν,Gǫ) with density p(σǫ,p) ∝ [1 + (σǫ,p/Gǫ)(2)/ν], σǫ,p > 0, where the hyperparameters ν and Gǫ are known constants (Gelman, 2006).”
Wang, Y., Xiao, Z., Tiong, R. L. K., & Zhang, L. (2021) · Applied Soft Computing · doi:10.1016/j.asoc.2021.107176 · Cited by 9
Bunji, K., & Okada, K. (2021) · Multivariate Behavioral Research · doi:10.1080/00273171.2021.1896351 · Cited by 9
Bayesian meta‐analysis using SAS PROC BGLIMM
Rott, K. W., Lin, L., Hodges, J. S., Siegel, L. K., Shi, A., Chen, Y., & Chu, H. (2021) · Research Synthesis Methods · doi:10.1002/jrsm.1513 · Cited by 8
A group‐sequential randomized trial design utilizing supplemental trial data
Kotalik, A., Vock, D. M., Hobbs, B. P., & Koopmeiners, J. S. (2021) · Statistics in Medicine · doi:10.1002/sim.9249 · Cited by 8
Trends in undernutrition mortality among children under five years of age and adults over 60
Uribe-Quintero, R., Castaño, L. S. Á., Caicedo-Velásquez, B., et al. (2022) · Biomédica · doi:10.7705/biomedica.5937 · Cited by 7
“0001 was assigned to the standard deviations σv and σu (20).”
MethodData-based priors for vector error correction models
Prüser, J. (2021) · International Journal of Forecasting · doi:10.1016/j.ijforecast.2021.10.007 · Cited by 8
Arrivederci! An Analysis of Tourism Impact in the Italian Provinces
Centinaio, A., Comerio, N., & Pacicco, F. (2022) · International Journal of Hospitality & Tourism Administration · doi:10.1080/15256480.2021.2025187 · Cited by 7
“With regard to the hierarchical BPVAR, we start our robustness analysis from the suggestion made by Jarociński (2010) and Gelman (2006) which indicates a value of s0 and v0 ¼ 0.001 for “neutral” priors when the number of cross-section units is higher than 5 (Dieppe et al., 2018), thus, for both…”
MethodInfluentialGeiger, M., Kistler, C., Mattmann, P., Jenni, L., Hegglin, D., & Bontadina, F. (2022) · Frontiers in Ecology and Evolution · doi:10.3389/fevo.2022.850442 · Cited by 8
“…informative prior distributions (Lemoine, 2019) for the intercepts and the variance parameters, i.e., student-t(3, −2.3,2.5) for the intercept of the count model, logistic (0, 1) for the intercept of the zero-model, and halfstudent-t(3, 0, 2.5) for the among-household variance (Gelman, 2006).”
BackgroundInfluentialZitzmann, S., Walther, J.-K., Hecht, M., & Nagengast, B. (2022) · Psych · doi:10.3390/psych4030029 · Cited by 7
Smith, M. J., Charlton, M. E., & Oleson, J. (2023) · Biometrical Journal · doi:10.1002/bimj.202200213 · Cited by 7
“We specify weakly informative prior distributions on all of the standard deviation terms, per the recommendation of Gelman (2006).”
MethodA Two-Component G-Prior for Variable Selection
Zhang, H., Huang, X., Gan, J., Karmaus, W., & Sabo‐Attwood, T. (2015) · Bayesian Analysis · doi:10.1214/15-ba953 · Cited by 7
A Bayesian Approach to Account for Misclassification and Overdispersion in Count Data
Wu, W., Stamey, J. D., & Kahle, D. J. (2015) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph120910648 · Cited by 7
“Gelman [17] find that both a half-Cauchy distribution and a uniform distribution perform better as priors than the inverse gamma.”
BackgroundA note on bimodality in the log-likelihood function for penalized spline mixed models
Welham, S. J., & Thompson, R. (2008) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2008.10.031 · Cited by 7
Simultaneous Confidence Intervals for Ratios of Fixed Effect Parameters in Linear Mixed Models
Schaarschmidt, F., & Djira, G. D. (2014) · Communications in Statistics - Simulation and Computation · doi:10.1080/03610918.2013.849741 · Cited by 7
“The particular choices of the distributional assumption and the parameters considered as noninformative may differ between authors (e.g., Browne and Draper, 2005; Gelman, 2006; Zhao et al., 2006), and of course the choice of the parameters depends on the scaling of the variables in the dataset.”
BackgroundSource distribution modelling for end‐member mixing in hydrology
Brewer, M., Tetzlaff, D., Malcolm, I. A., et al. (2011) · Environmetrics · doi:10.1002/env.1110 · Cited by 7
“We complete the model definition by defining appropriate diffuse priors:Nð0; 10Þ for the overall mean parametermq and for the linear term v in Equation (7); and Uniform over a suitable range for the standard deviation parameters sq, sh and su (as recommended by Gelman, 2006).”
MethodDose finding with continuous outcome in phase I oncology trials
Wang, Y., & Ivanova, A. (2014) · Pharmaceutical Statistics · doi:10.1002/pst.1662 · Cited by 7
Bayesian analysis of spatial data using different variance and neighbourhood structures
Rampaso, R. C., de Souza, A. D. P., & Flores, E. F. (2015) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2015.1022549 · Cited by 14
“Following the work of Gelman,[22] to the standard deviation was specified a non-informative prior with uniform distribution, p(σ ) ∼ U(0, T).”
MethodZhang, J., Bailer, A. J., & Oris, J. T. (2012) · Environmental Toxicology and Chemistry · doi:10.1002/etc.1886 · Cited by 7
Hierarchical Metapopulation Dynamics of Two Aphid Species on a Shared Host Plant
Zheng, C., Weisser, W. W., Härri, S. A., et al. (2009) · The American Naturalist · doi:10.1086/603623 · Cited by 7
“Because the prior used for the variance compo-{C, E} nents is likely to influence the results (Gelman 2006), we tested the sensitivity of our results to this choice of the prior by setting the scale parameter of the half-Cauchy distribution to 0.25 or to 1.0.”
MethodInfluentialPredicting county‐level cancer incidence rates and counts in the USA
Yu, B. (2013) · Statistics in Medicine · doi:10.1002/sim.5833 · Cited by 7
Bayesian spatio-temporal random coefficient time series (BaST-RCTS) model of infectious disease
Zhang, T., Zhang, X., Ma, Y., Zhou, X., Liu, Y., Feng, Z., & Li, X. (2014) · Mathematical Biosciences · doi:10.1016/j.mbs.2014.09.015 · Cited by 7
Metadata
- Authors
- Gelman, A.
- Affiliations
- Columbia University (US)
- Corresponding author
- Gelman, Andrew
- ORCID
- Gelman, Andrew
- Venue
- Bayesian Analysis
- Publisher
- Institute of Mathematical Statistics
- Published
- 2006-09-01
- Volume
- 1
- Issue
- 3
- Pages
- 515-534
- ISSN
- 1931-6690 (ISSN-L)
- 1936-0975
- OpenAlex
- W1536497620
- Semantic Scholar
- 63d40805d4635b70f6b5aa07fc2c9821480cfc1e
- Cited by
- 4,050
- References
- 34
- Citations by year
- 2026: 194 · 2025: 218 · 2024: 237 · 2023: 237 · 2022: 282 · 2021: 312 · 2020: 320 · 2019: 281 · 2018: 266 · 2017: 263 · 2016: 235 · 2015: 217 · 2014: 176 · 2013: 142 · 2012: 184
- FWCI
- 78.70
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- cc-by
- Locations
- Bayesian Analysis (publishedVersion, cc-by)
- Project Euclid (Cornell University) (submittedVersion)
- Funders
- National Science Foundation
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods and Bayesian Inference
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Distribution Estimation and Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Bayesian Methods and Mixture Models
- Keywords
- prior distributions, hierarchical models, scale parameter, non-informative priors, weakly informative priors, uniform prior, analysis of variance
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-02, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-02 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.