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 1,601–1,700 of 4,089
Staton, B. A., Catalano, M. J., & Fleischman, S. J. (2016) · Fisheries Research · doi:10.1016/j.fishres.2016.09.001 · Cited by 22
Martin, J. S., & Suarez, S. A. (2017) · American Journal of Primatology · doi:10.1002/ajp.22670 · Cited by 16
“A parameter expanded prior was utilized to enhance MCMC mixing properties and weakly inform (co)variance parameter estimates (Gelman, 2006).”
MethodSpatiotemporal multivariate mixture models for Bayesian model selection in disease mapping
Lawson, A. B., Carroll, R. M., Faes, C., et al. (2017) · Environmetrics · doi:10.1002/env.2465 · Cited by 14
“It was found that C= 4 was essentially noninformative through a sensitivity analysis that included comparisons to uniform distributions with larger ranges (Gelman, 2006).”
BackgroundLittle, D., Cheng, H. H., & Wright, B. A. (2018) · Attention Perception & Psychophysics · doi:10.3758/s13414-018-1584-x · Cited by 14
“The hyperpriors in the multilevel regression were weakly informative Cauchy distributions (scale 1) (Gelman, 2006).”
MethodChevalier, M., & Knape, J. (2019) · Oikos · doi:10.1111/oik.06401 · Cited by 15
Li, J., Zhang, X., Wang, L., et al. (2019) · BMC Public Health · doi:10.1186/s12889-019-7758-1 · Cited by 14
“Overall temporal trend was described by (b0t * + vt), vt ~ N (0, σ(2)v ) [28], and t * t=t − tmid indicates the temporal span.”
BackgroundMatthews, J., Nagao, K. J., Ding, C., Newby, R., Kempster, P. A., & Hohwy, J. (2020) · Cortex · doi:10.1016/j.cortex.2019.12.009 · Cited by 14
Evaluating the informativeness of the Besag-York-Mollié CAR model
Quick, H., Song, G., & Tabb, L. P. (2021) · Spatial and Spatio-temporal Epidemiology · doi:10.1016/j.sste.2021.100420 · Cited by 14
“Furthermore, while much research has been done to construct weakly informative priors (Gelman, 2006) or to theoretically derive prior distributions that penalize complexity (Simpson et al., 2017), the contribution of this paper is to quantify how informative the model is for certain values of σ2…”
BackgroundCagnazzi, D., Parra, G. J., Harrison, P. L., Brooks, L. O., & Rankin, R. (2020) · Global Ecology and Conservation · doi:10.1016/j.gecco.2020.e01203 · Cited by 13
Bayesian Semiparametric Longitudinal Drift-Diffusion Mixed Models for Tone Learning in Adults
Paulon, G., Llanos, F., Chandrasekaran, B., & Sarkar, A. (2020) · Journal of the American Statistical Association · doi:10.1080/01621459.2020.1801448 · Cited by 16
“We let σ 2β ,1 ∼ C+(0, 1), where C+(a, b) denotes a half-Cauchy distribution (Gelman 2006; Polson and Scott 2012) with location parameter a and scale parameter b.”
BackgroundHow vague is vague? How informative is informative? Reference analysis for Bayesian meta‐analysis
Ott, M., Plummer, M., & Roos, M. (2021) · Statistics in Medicine · doi:10.1002/sim.9076 · Cited by 13
Robinson, T. J., Pintar, A. L., Anderson‐Cook, C. M., & Hamada, M. S. (2012) · Journal of Quality Technology · doi:10.1080/00224065.2012.11917903 · Cited by 15
“For , we use the Uniform(0, 100) di↵use prior distribution as suggested by Gelman (2006).”
MethodStevenson, B. C., & Millar, R. B. (2013) · Environmental and Ecological Statistics · doi:10.1007/s10651-013-0236-5 · Cited by 15
“In particular, the prior on τ had a lower bound at 1, and the standard deviations of the random effects, σ f and σd , were given flat priors on the positive reals (Gelman 2006).”
BackgroundUsing spline models to estimate the varying health risks from air pollution across Scotland
Lee, D. (2012) · Statistics in Medicine · doi:10.1002/sim.5420 · Cited by 14
Causal Falling Rule ListsPreprint
Wang, F., & Rudin, C. D. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1510.05189 · Cited by 15
Fabrizi, E., Ferrante, M. R., & Trivisano, C. (2016) · Wiley eBooks · doi:10.1002/9781118814963.ch16 · Cited by 17
bayesmeta: Bayesian Random-Effects Meta-Analysis and Meta-Regression
Röver, C. (2015) · doi:10.32614/cran.package.bayesmeta · Cited by 14
Crowdsourcing prior information to improve study design and data analysis
Chrabaszcz, J. S., Tidwell, J. W., & Dougherty, M. R. P. (2017) · PLoS ONE · doi:10.1371/journal.pone.0188246 · Cited by 13
Efficacy of depletion models for estimating abundance of endangered fishes in streams
Stewart, D. R., Butler, M. J., Johnson, L. A., et al. (2018) · Fisheries Research · doi:10.1016/j.fishres.2018.09.026 · Cited by 13
Power link functions in an ordinal regression model with Gaussian process priors
Li, D., Wang, X., & Dey, D. K. (2019) · Environmetrics · doi:10.1002/env.2564 · Cited by 13
“…random effects f follow a multivariate normal distribution with the covariance matrix depending on the scale hyperparameter 𝜎2 and the length-scale hyperparameter vector 𝝆, which are all given weakly informative half-Cauchy priors with a small scale (Gelman, 2006), that is, half-Cauchy(0, 2.5).”
MethodInfluentialFuentes-González, J. A., Polly, P. D., & Martins, E. P. (2020) · Evolution · doi:10.1111/evo.13899 · Cited by 15
“It is common in Bayesian analysis to parameterize variances in terms of precision, so a uniform distribution can be applied on the inverse of the phenotypic variance to specify a diffuse prior (Gelman 2006): γ−1 U (0, 100).”
MethodBayesian network mediation analysis with application to the brain functional connectome
Zhao, Y., Chen, T., Cai, J., Lichenstein, S. D., Potenza, M. N., & Yip, S. W. (2022) · Statistics in Medicine · doi:10.1002/sim.9488 · Cited by 18
Thạch, N. N. (2023) · SAGE Open · doi:10.1177/21582440231181540 · Cited by 16
“According to Bayesians (Gelman, 2006; D. M. McNeish, 2016; D. McNeish & Stapleton, 2016; van de Schoot et al., 2015), variance terms are often problematic, so we set Inverse Gamma distributions of (0.01, 0.01) (non-informative) and (0.5, 0.5) (strongly informative) for the overall variance and…”
MethodInfluentialA Spatial Dynamic Factor Repeat Sales Model for Real Estate
Francke, M., Rolheiser, L., & de Minne, A. V. (2022) · The Journal of Real Estate Finance and Economics · doi:10.1007/s11146-023-09957-w · Cited by 14
Zou, Q., & Wen, J. (2023) · International Journal of Fatigue · doi:10.1016/j.ijfatigue.2023.107955 · Cited by 14
Prevalence of micronutrient deficiencies across diverse environments in rural Madagascar
Golden, C. D. W., Zamborain‐Mason, J., Levis, A., Rice, B. L., Allen, L. H., Hampel, D., Hazen, J., Metcalf, C. J. E., Randriamady, H. J., Shahab‐Ferdows, S., et al. (2024) · Frontiers in Nutrition · doi:10.3389/fnut.2024.1389080 · Cited by 22
Coventry, B. S., & Bartlett, E. L. (2024) · eNeuro · doi:10.1523/eneuro.0484-23.2024 · Cited by 19
“A half-Cauchy distribution is once again chosen as it is weakly informative and allows for extreme values if data dictate (Gelman, 2006; Polson and Scott, 2012).”
Loyalty Currency and Mental Accounting: Do Consumers Treat Points Like Money?
Lim, F., Chun, S. Y., & Satopää, V. (2021) · Manufacturing & Service Operations Management · doi:10.1287/msom.2022.0287 · Cited by 19
Predictive Modeling of Expressed Emotions in Music Using Pairwise Comparisons
Madsen, J., Jensen, B. S., & Larsen, J. O. (2013) · Lecture notes in computer science · doi:10.1007/978-3-642-41248-6_14 · Cited by 15
Modeling Age and Nest‐Specific Survival Using a Hierarchical Bayesian Approach
Cao, J., He, C. Z., Wells, K. M. S., et al. (2009) · Biometrics · doi:10.1111/j.1541-0420.2009.01204.x · Cited by 13
“We have also considered a uniform prior on√ τi , [τi ] ∝ 1/ √ τi , which is recommended by Gelman (2006).”
Statistical methods to correct for observation error in a density‐independent population model
Buonaccorsi, J. P., & STAUDENMAYER, J. W. (2009) · Ecological Monographs · doi:10.1890/07-1299.1 · Cited by 13
Wandrag, E. M., SHEPPARD, A. W., Duncan, R. P., et al. (2013) · Perspectives in Plant Ecology Evolution and Systematics · doi:10.1016/j.ppees.2013.03.002 · Cited by 13
Mutsvari, T., Bandyopadhyay, D., Declerck, D., & Lesaffre, E. M. E. H. (2013) · Statistics in Medicine · doi:10.1002/sim.5944 · Cited by 13
“0; 100/ priors were assumed for the standard deviations m and tm of the random effects [24].”
BackgroundGenomic Prediction Accounting for Residual Heteroskedasticity
Ou, Z., Tempelman, R. J., Steibel, J. P., Ernst, C. W., Bates, R. O., & Bello, N. M. (2015) · G3 Genes Genomes Genetics · doi:10.1534/g3.115.022897 · Cited by 14
Avoiding Boundary Estimates in Linear Mixed Models Through Weakly Informative Priors
Chung, Y., Rabe‐Hesketh, S., Gelman, A., Liu, J., & Dorie, V. (2012) · Collection of Biostatistics Research Archive · Cited by 17
Self-citationLow-Rank Independence Samplers in Hierarchical Bayesian Inverse Problems
Brown, D. A., Saibaba, A. K., & Vallélian, S. (2018) · SIAM/ASA Journal on Uncertainty Quantification · doi:10.1137/17m1137218 · Cited by 12
Scroggie, M. P., Forsyth, D. M., McPhee, S. R., Matthews, J., Stuart, I., Stamation, K. A., Lindeman, M., & Ramsey, D. S. L. (2018) · Journal of Applied Ecology · doi:10.1111/1365-2664.13253 · Cited by 14
“Weakly informative half−t4 priors (Gelman, 2006) were specified for the SDs of the process and sampling errors and for transectlevel random effects.”
Inter-industry wage inequality: persistent differences and turbulent equalisation
Mokre, P., & Rehm, M. (2020) · Cambridge Journal of Economics · doi:10.1093/cje/bez064 · Cited by 17
Więcek, W., Dorne, J. L. C. M., Quignot, N., Béchaux, C., & Amzal, B. (2019) · Computational Toxicology · doi:10.1016/j.comtox.2019.100106 · Cited by 12
Mullah, M. A. S., Hanley, J. A., & Benedetti, A. (2019) · BMC Medical Research Methodology · doi:10.1186/s12874-019-0861-2 · Cited by 13
Inferring the Allelic Series at QTL in Multiparental Populations
Crouse, W. L., Kelada, S. N. P., & Valdar, W. S. J. (2020) · Genetics · doi:10.1534/genetics.120.303393 · Cited by 12
“The half-t is a preferred prior choice for variance components in normally distributedmodels due to its behavior at the boundary ðf ¼ 0Þ and its convenient representation as the product of two conditionally conjugate latent variables (Gelman 2006).”
BackgroundA Hybrid Gibbs Sampler for Edge-Preserving Tomographic Reconstruction with Uncertain View Angles
Uribe, F., Bardsley, J. M., Dong, Y., Hansen, P. C., & Riis, N. A. B. (2022) · SIAM/ASA Journal on Uncertainty Quantification · doi:10.1137/21m1412268 · Cited by 22
Understanding the Deviance Information Criterion for SEM: Cautions in Prior Specification
Liu, H., Depaoli, S. A., & Marvin, L. (2021) · Structural Equation Modeling A Multidisciplinary Journal · doi:10.1080/10705511.2021.1994407 · Cited by 15
“The amount of prior information that is incorporated differs across different sets of priors for a given model (Depaoli, 2013; Depaoli et al., 2021; Gelman, 2006; Liu et al., 2016).”
BackgroundMaior, C. B. S., Macêdo, J. B., Lins, I. D., Moura, M. D. C., Azevedo, R. V., Santana, J. M., da Silva, M. F., & Magalhães, M. V. (2021) · Journal of Petroleum Science and Engineering · doi:10.1016/j.petrol.2021.109891 · Cited by 20
Sugimoto, N., Fukasawa, K., Asahara, A., Kasada, M., Matsuba, M., & Miyashita, T. (2022) · Proceedings of the Royal Society B Biological Sciences · doi:10.1098/rspb.2021.2222 · Cited by 13
The Evolution of Religion and Morality project: reflections and looking ahead
Purzycki, B., Lang, M., Henrich, J., & Norenzayan, A. (2022) · Religion, Brain & Behavior · doi:10.1080/2153599x.2021.2021546 · Cited by 8
“The burden of effort, however, lies in carefully defining and examining one’s prior distributions of intercepts and, if necessary, varying effects (see Gelman, 2006).”
BackgroundFirst, Do No Harm: Predictive Analytics to Reduce In-Hospital Adverse Events
Lin, Y.-K., & Fang, X. (2021) · Journal of Management Information Systems · doi:10.1080/07421222.2021.1990619 · Cited by 22
Addy, J., Ellis, R. H., MacLaren, C., Macdonald, A. J., Semenov, M. A., & Mead, A. (2022) · Journal of The Royal Society Interface · doi:10.1098/rsif.2022.0361 · Cited by 13
Jaya, I. G. N. M., Folmer, H., & Lundberg, J. (2022) · The Annals of Regional Science · doi:10.1007/s00168-022-01191-1 · Cited by 17
Bayesian space-time SIR modeling of Covid-19 in two US states during the 2020–2021 pandemic
Lawson, A. B., & Kim, J. (2022) · PLoS ONE · doi:10.1371/journal.pone.0278515 · Cited by 17
“1)) which is reasonably noninformative but prohibits large precisions [11].”
BackgroundTime since fire shapes plant immaturity risk across fire severity classes
Plumanns-Pouton, E. S., Swan, M. H., Penman, T. D., Collins, L., & Kelly, L. T. (2023) · Fire Ecology · doi:10.1186/s42408-023-00185-4 · Cited by 17
Larval settlement in benthic environments: the effects of velocity and bed element geometry
Fingerut, J. T., Hart, D. D., & Thomson, J. R. (2010) · Freshwater Biology · doi:10.1111/j.1365-2427.2010.02535.x · Cited by 13
Carpenter, J. A., Hadfield, J. D., Bangham, J., et al. (2011) · Evolution · doi:10.1111/j.1558-5646.2011.01501.x · Cited by 13
Assessing the predictive ability of a multilevel binary regression model
Oirbeek, R. V., & Lesaffre, E. M. E. H. (2011) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2011.11.023 · Cited by 13
Li, J., Gray, B. R., & Bates, D. M. (2008) · Communications in Statistics - Simulation and Computation · doi:10.1080/03610910802361366 · Cited by 13
“Gelman ( 2006 ) brought out important issues on the prior choices of variance parameters in multi-level linear models.”
BackgroundBandyopadhyay, D., Castro, L. M., Lachos, V. H., & Pinheiro, H. P. (2015) · Journal of Agricultural Biological and Environmental Statistics · doi:10.1007/s13253-014-0195-9 · Cited by 13
Estimating tree growth from complex forest monitoring data
Eitzel, M. V., Battles, J. J., York, R. A., Knape, J., & de Valpine, P. (2013) · Ecological Applications · doi:10.1890/12-0504.1 · Cited by 16
“Priors allowing the standard deviation to approach zero, including uniform and half-t-distribution priors (Gelman 2006, Knape et al. 2008), result in bad mixing as the system struggles to move away from a scenario in which observations perfectly match latent states.”
BackgroundYoung, J., Potter, M., Cox, J., et al. (2013) · CMAJ Open · doi:10.9778/cmajo.20130009 · Cited by 13
Hemming, K., Chilton, P. J., Lilford, R. J., Avery, A. J., & Sheikh, A. (2012) · PLoS ONE · doi:10.1371/journal.pone.0038306 · Cited by 14
Wu, J., Liu, Y., & Zhou, S. (2016) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2016.2566579 · Cited by 12
“However, Gelman [26] showed that the inferences become very sensitive to ω for data sets in which low values of random effects variance are possible, and the prior distribution hardly looks noninformative.”
BackgroundInfluentialWolf, L. J., Anselin, L., & Arribas‐Bel, D. (2017) · Geographical Analysis · doi:10.1111/gean.12135 · Cited by 14
A Bayesian Meta-analysis Method for Estimating Risk Difference of Rare Events
Tang, Y., Tang, Q., Yu, Y., & Wen, S. (2017) · Journal of Biopharmaceutical Statistics · doi:10.1080/10543406.2017.1372767 · Cited by 14
Rousselière, D. (2019) · Journal of Quantitative Economics · doi:10.1007/s40953-019-00164-0 · Cited by 15
“We used weakly informative priors (Independent Cauchy prior distribution with center 0 and scale 2.5 on population parameters and a half Student-t prior with 3 degrees of freedom on the group level parameter), following Gelman (2006) and Buerkner (2017).”
MethodInfluentialEvidence synthesis from aggregate recurrent event data for clinical trial design and analysis
Holzhauer, B., Wang, C., & Schmidli, H. (2017) · Statistics in Medicine · doi:10.1002/sim.7549 · Cited by 12
Low‐Choy, S., Riley, T., & Alston‐Knox, C. (2017) · Educational Media International · doi:10.1080/09523987.2017.1397404 · Cited by 13
“Priors for the variance components were truncated Normals on the scale of the standard deviations, as per Gelman (2006).”
BackgroundMielke, J., Schmidli, H., & Jones, B. C. (2018) · Biometrical Journal · doi:10.1002/bimj.201700152 · Cited by 13
“As priors for the betweentrial standard deviation in hierarchical models, half-normal or half-t distributions are recommended; see Spiegelhalter, Abrams, and Myles (2004) and Gelman (2006).”
Bolin, J. H., Finch, W. H., & Stenger, R. (2018) · Educational and Psychological Measurement · doi:10.1177/0013164418773494 · Cited by 13
“Bayesian estimation using the MCMC estimator has been suggested as being advantageous for small sample cases, including in the context of multilevel data (Gelman, 2006).”
BackgroundMethodPopularity Tracking for Proactive Content Caching with Dynamic Factor Analysis
Mehrizi, S., Tsakmalis, A., Shahbazpanahi, S., et al. (2019) · IEEE/CIC International Conference on Communications in China (ICCC) · doi:10.1109/iccchina.2019.8855852 · Cited by 13
“Similar to [17], we use the heavy-tailed half-Cauchy distribution with scale parameter A as a prior for the variance of each column σk ∼ C (0, A).”
MethodBayesian Joint Matrix Decomposition for Data Integration with Heterogeneous Noise
Zhang, C., & Zhang, S. (2019) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2019.2946370 · Cited by 12
Sahoo, M. M., & Patra, K. C. (2020) · African Journal of Science Technology Innovation and Development · doi:10.1080/20421338.2019.1692460 · Cited by 13
Assessment and Management of Small Yellow Croaker (Larimichthys polyactis) Stocks in South Korea
Choi, M., & Kim, D.-H. (2020) · Sustainability · doi:10.3390/su12198257 · Cited by 13
The horseshoe prior for time-varying parameter VARs and Monetary Policy
Prüser, J. (2021) · Journal of Economic Dynamics and Control · doi:10.1016/j.jedc.2021.104188 · Cited by 22
Constructing Representative Air Quality Indicators with Measures of Uncertainty
Lee, D., Ferguson, C. A., & Scott, E. M. (2010) · Journal of the Royal Statistical Society Series A (Statistics in Society) · doi:10.1111/j.1467-985x.2010.00658.x · Cited by 13
“A flat prior on the standard deviation scale is specified for σ2 as suggested by Gelman (2006), because the commonly used inverse-gamma.0:01, 0:01/ prior has very little mass close to zero, resulting in an upward bias if the true value of σ2 is small.”
MethodRejoinder to commentaries on ‘The BUGS project: Evolution, critique and future directions’
Lunn, D. J., Spiegelhalter, D. J., Thomas, A., et al. (2009) · Statistics in Medicine · doi:10.1002/sim.3691 · Cited by 13
Evaluation of the Accuracy of Diagnostic Tests From Repeated Measurements Without a Gold Standard
ENGEL, B., Backer, J. A., & Buist, W. G. (2010) · Journal of Agricultural Biological and Environmental Statistics · doi:10.1007/s13253-009-0013-y · Cited by 13
“Choosing ±log(6) for the fifth and 95th percentiles ±1.645√2σ1, an upper bound for a Uniform prior (Gelman 2006) for σ1 (and σ0) of 0.77 was obtained.”
MethodInfluentialHosack, G. R., Trenkel, V. M., & Dambacher, J. M. (2012) · Journal of Marine Systems · doi:10.1016/j.jmarsys.2012.09.003 · Cited by 13
“…fluctuates by at least ∼ 10%).307 Given the above constraints on the standard deviation of the noise pro-308 cesses and based on arguments presented by Gelman (2006), we place uniform309 priors on the standard deviation of the process and observation noise with310 limits determined by the bounds…”
BackgroundMethodA smoothed ANOVA model for multivariate ecological regression
Marí-Dell’Olmo, M., Martínez‐Beneito, M. A., Gotsens, M., et al. (2013) · Stochastic Environmental Research and Risk Assessment · doi:10.1007/s00477-013-0782-2 · Cited by 13
“A half-normal distribution with mean 0 and precision 0.0001 was assigned to the standard deviations rf1 ; rf2 ; rU1; rU2; rV1 and rV2 (Gelman 2006).”
MethodInfluentialSparse variational analysis of linear mixed models for large data sets
Armagan, A., & Dunson, D. B. (2011) · Statistics & Probability Letters · doi:10.1016/j.spl.2011.02.029 · Cited by 13
“Although the parameterization is redundant, it has been noted that the redundant parameterizations are often useful for computational reasons and for inducing new classes of priors with appealing properties (Gelman, 2006).”
BackgroundInterweaving Markov Chain Monte Carlo Strategies for Efficient Estimation of Dynamic Linear Models
Simpson, M. J., Niemi, J., & Roy, V. (2015) · Journal of Computational and Graphical Statistics · doi:10.1080/10618600.2015.1105748 · Cited by 11
“In addition, there are well-known inferential problems with the inverse Wishart prior in the hierarchical model literature, for example, Gelman (2006).”
Improved Laplace approximation for marginal likelihoods
Ruli, E., Sartori, N., & Ventura, L. (2016) · Electronic Journal of Statistics · doi:10.1214/16-ejs1218 · Cited by 13
Lobo, M. S., & Yao, D. (2010) · SSRN Electronic Journal · doi:10.2139/ssrn.1638811 · Cited by 13
Baker, C. C. M., Ji, Y., Popescu, V. D., Wang, J., Wu, C., Wang, Z., Li, Y., Wang, L., Hua, C., Zhong-xing, Y., et al. (2020) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2020.02.10.941336 · Cited by 22
The impact of prior information on estimates of disease transmissibility using Bayesian tools
Moser, C. B., Gupta, M., Archer, B. N., & White, L. F. (2015) · PLOS ONE · doi:10.1371/journal.pone.0118762 · Cited by 11
McDonald, P., Brown, H. A., Topham, T. H., Kelly, M. K., Jardine, W. T., Carr, A. J., Sawka, M. N., Woodward, A. P., Clark, B., & Périard, J. D. (2025) · Comprehensive physiology · doi:10.1002/cph4.70017 · Cited by 24
“Scaled priors were applied to all continuous predictor variables (Gelman 2006).”
Bates, A., Laven, R. A., O’Brien, R., Liggett, S., & Griffin, F. (2020) · Preventive Veterinary Medicine · doi:10.1016/j.prevetmed.2020.105199 · Cited by 14
Bayesian Models for N-of-1 Trials
Schmid, C. H., & Yang, J. (2022) · Harvard Data Science Review · doi:10.1162/99608f92.3f1772ce · Cited by 22
“Common choices are uniform distributions ranging between zero and an upper bound or a folded distribution such as a half-normal or half-t that only take positive values (Gelman, 2006)(Röver et al., 2021).”
BackgroundDeveloping an empirical model for spillover and emergence: Orsay virus host range in Caenorhabditis
Shaw, C. L., & Kennedy, D. A. (2022) · Proceedings of the Royal Society B Biological Sciences · doi:10.1098/rspb.2022.1165 · Cited by 12
Goplerud, M. (2023) · American Political Science Review · doi:10.1017/s0003055423000035 · Cited by 16
Bayesian spatial modelling of tuberculosis-HIV co-infection in Ethiopia
Gemechu, L. L., & Debusho, L. K. (2023) · PLoS ONE · doi:10.1371/journal.pone.0283334 · Cited by 15
Daniells, L., Mozgunov, P., Bedding, A. W., & Jaki, T. (2023) · Statistics in Medicine · doi:10.1002/sim.9867 · Cited by 12
Monnahan, C. C. (2024) · Fisheries Research · doi:10.1016/j.fishres.2024.107024 · Cited by 19
Newell, S. E., Doll, J. C., Jutte, M. C., et al. (2024) · Harmful Algae · doi:10.1016/j.hal.2024.102684 · Cited by 12
Avnimelech, G., Dushnitsky, G., Ellsaesser, F., & Fitza, M. (2024) · Strategic Management Journal · doi:10.1002/smj.3665 · Cited by 19
Forest Fragmentation and Warmer Climate Increase Tick-Borne Disease Infection
Iijima, H., Watari, Y., Doi, K., Yasuo, K., & Okabe, K. (2025) · EcoHealth · doi:10.1007/s10393-025-01702-4 · Cited by 14
“Prior distributions for a and all b s in the model were set as vague normal distributions [N(0, 10 4 )], whereas the prior distribution for r was a vague uniform distribution [U(0, 100)] (Gelman 2006).”
Evaluating the analytical performance of direct-to-consumer gut microbiome testing services
Servetas, S. L., Gierz, K., Hoffmann, D., Ravel, J., & Jackson, S. A. (2026) · Communications Biology · doi:10.1038/s42003-025-09301-3 · Cited by 18
“1 quantitatively, we can perform an inferential analysis on σ 2 a = σ 2 e using a hierarchical Bayesian model with priors that are generally noncommittal, but stay within the known restrictions of the data 40 – 42 .”
Selection for increased body length in Subantarctic fur seals on Amsterdam Island
Authier, M., Cam, E., & Guinet, C. (2010) · Journal of Evolutionary Biology · doi:10.1111/j.1420-9101.2010.02193.x · Cited by 12
“Weakly informative Cauchy priors (Gelman, 2006; Gelman et al., 2008) were used.”
MethodRegional Differentiation and Farm Exit: A Hierarchical Model for Tuscany
Landi, C., Stefani, G., Rocchi, B., et al. (2015) · Journal of Agricultural Economics · doi:10.1111/1477-9552.12130 · Cited by 12
A Bayesian Approach to Fitting Gibbs Processes with Temporal Random Effects
King, R., Illian, J., KING, S. E., Nightingale, G., & Hendrichsen, D. K. (2012) · Journal of Agricultural Biological and Environmental Statistics · doi:10.1007/s13253-012-0111-0 · Cited by 15
“In particular, we increase the prior standard deviation of the Normal priors on the underlying intensity rate, θ 1 , and interaction parameters, θ 2 , by a factor of 10 and also specify σ ∼ U [ 0 , 100 ] (Gelman 2006).”
BackgroundJoint Bayesian modeling of time to malaria and mosquito abundance in Ethiopia
Belay, D. B., Kifle, Y. G., Goshu, A. T., Gran, J. M., Yewhalaw, D., Duchateau, L., & Frigessi, A. (2017) · BMC Infectious Diseases · doi:10.1186/s12879-017-2496-4 · Cited by 12
New methods for small area estimation with linkage uncertainty
Briscolini, D., Consiglio, L. D., Liseo, B., et al. (2018) · International Journal of Approximate Reasoning · doi:10.1016/j.ijar.2017.12.005 · Cited by 12
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.