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 1,301–1,400 of 4,089
Martins, R., Silva, G. L., & Andreozzi, V. (2017) · Biometrical Journal · doi:10.1002/bimj.201600159 · Cited by 19
“Gelman (2006) discusses these and other hyperprior issues proposing several alternatives such as half-normal, half-Cauchy, or (proper) uniform priors for the standard deviation. Recently, Klein and Kneib (2016) developed the idea of scale dependent priors for the variance parameters in structured additive distribution…”
BackgroundThomas, F., & Vesk, P. A. (2017) · PLoS ONE · doi:10.1371/journal.pone.0176959 · Cited by 19
N-mix for fish: estimating riverine salmonid habitat selection via N-mixture models
Som, N. A., Perry, R. W., Jones, E. C., et al. (2017) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/cjfas-2017-0027 · Cited by 19
“4), we specified uniform (a, b) priors (Gelman 2006), with a = 0 and b = 10.”
MethodIijima, H., & Otsu, C. (2018) · Biodiversity and Conservation · doi:10.1007/s10531-018-1577-z · Cited by 21
Bayesian methods for type A evaluation of standard uncertainty
van der Veen, A. M. H. (2018) · Metrologia · doi:10.1088/1681-7575/aad103 · Cited by 25
Heterogeneous tail generalized COMFORT modeling via Cholesky decomposition
Näf, J., Paolella, M. S., & Polak, P. (2019) · Journal of Multivariate Analysis · doi:10.1016/j.jmva.2019.02.004 · Cited by 17
Cyclists injured in traffic crashes in Hong Kong: A call for action
Xu, P., Dong, N., Wong, S. C., & Huang, H. (2019) · PLoS ONE · doi:10.1371/journal.pone.0220785 · Cited by 23
Bayesian factor analytic model: An approach in multiple environment trials
Nuvunga, J. J., da Silva, C. P., de Oliveira, L. A., de Lima, R. R., & Balestre, M. (2019) · PLoS ONE · doi:10.1371/journal.pone.0220290 · Cited by 20
Heterogeneity of Variances in the Bayesian AMMI Model for Multienvironment Trial Studies
da Silva, C. P., de Oliveira, L. A., Nuvunga, J. J., Pamplona, A. K. A., & Balestre, M. (2019) · Crop Science · doi:10.2135/cropsci2018.10.0641 · Cited by 19
Kim, S., Huh, D., Zhou, Z., & Mun, E. (2020) · International Journal of Behavioral Development · doi:10.1177/0165025419894730 · Cited by 19
I can see myself helping: The effect of self‐awareness on prosocial behaviour
Lewis, J. A., Himmelberger, Z. M., & Elmore, J. D. (2020) · International Journal of Psychology · doi:10.1002/ijop.12733 · Cited by 20
Li, Q., Bedi, T., Lehmann, C. U., Xiao, G., & Xie, Y. (2021) · GigaScience · doi:10.1093/gigascience/giab009 · Cited by 21
“We recommend small values, such as a φ = b φ = 0.001, for a non-informative setting [48].”
InfluentialVariational inference for sparse spectrum Gaussian process regression
Tan, L. S. L., Ong, V. M.-H., Nott, D. J., & Jasra, A. (2015) · Statistics and Computing · doi:10.1007/s11222-015-9600-7 · Cited by 19
“We set ρ = 1.5, Aσ = Aγ = 25 for the halfCauchy priors, following Gelman (2006) and Wand et al. (2011) and µ0λ = 0, Σ 0 λ = 10Id for the lengthscales in Algorithm 2.”
MethodInfluentialSatterthwaite, W. H., Mohr, M. S., O’Farrell, M. R., & Wells, B. K. (2012) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/f2012-036 · Cited by 23
Hierarchical Latency Models for Dose-Time-Response Associations
Richardson, D. B., MacLehose, R. F., Langholz, B. M., & Cole, S. R. (2011) · American Journal of Epidemiology · doi:10.1093/aje/kwq387 · Cited by 23
Alameddine, I., Karmakar, S., Qian, S. S., Paerl, H. W., & Reckhow, K. H. (2013) · Water Resources Research · doi:10.1002/wrcr.20509 · Cited by 20
Anders, R., Alario, F., & Batchelder, W. H. (2018) · Cross-Cultural Research · doi:10.1177/1069397117727500 · Cited by 19
“In this way, the approach has been found to improve estimation for smaller sample sizes by constraining error and maximizing information that can be gained from the data (see Gelman, 2006; Kruschke, 2011; Lee, 2011; Rouder, Morey, & Pratte, 2015).”
MethodLele, S. R. (2020) · Frontiers in Ecology and Evolution · doi:10.3389/fevo.2019.00501 · Cited by 25
“Gelman (2006) discusses similar computational problems associated with the non-informative priors for variance components and concludes uniform prior is, in fact, a good choice and not the commonly used inverse Gamma prior (e.g., King et al., 2009).”
MethodReligious belief and cognitive conflict sensitivity: A preregistered fMRI study
Hoogeveen, S., Snoek, L., & van Elk, M. (2020) · Cortex · doi:10.1016/j.cortex.2020.04.011 · Cited by 22
Bayesian network meta-analysis for unordered categorical outcomes with incomplete data
Schmid, C. H., Trikalinos, T. A., & Olkin, I. (2013) · Research Synthesis Methods · doi:10.1002/jrsm.1103 · Cited by 20
“Gelman (2006) and Lambert et al. (2005) discuss choosing noninformative prior distributions, mainly in the context of vague prior distributions for the random effects variance in the univariate setting.”
BackgroundLiu, L., Liu, D., Guo, T., Schwieter, J. W., & Liu, H. (2023) · NeuroImage · doi:10.1016/j.neuroimage.2023.120393 · Cited by 18
“deviations, we used Cauchy prior distributions, which tends to give sharper and more reasonable estimates than uniform or inverse-Gaussian prior distributions (Gelman, 2006).”
MethodEfficient Marginalization-Based MCMC Methods for Hierarchical Bayesian Inverse Problems
Saibaba, A. K., Bardsley, J. M., Brown, D. A., & Alexanderian, A. (2019) · SIAM/ASA Journal on Uncertainty Quantification · doi:10.1137/18m1220625 · Cited by 17
“( Other choices are also possible and are discussed in [15, 26, 7] and elsewhere.”
BackgroundŻelazny, W. R., & Lukáš, J. (2020) · Remote Sensing · doi:10.3390/rs12203462 · Cited by 21
“Because of the variety of the indexes, the modelling assumed uninformative priors [55,84].”
BackgroundHsieh, N., Bois, F. Y., Tsakalozou, E., Ni, Z., Yoon, M., Sun, W., Klein, M., Reisfeld, B., & Chiu, W. A. (2021) · Journal of Pharmacokinetics and Pharmacodynamics · doi:10.1007/s10928-021-09778-5 · Cited by 18
White, G. W., McConville, K. S., Moisen, G. G., et al. (2021) · Frontiers in Forests and Global Change · doi:10.3389/ffgc.2021.752911 · Cited by 18
“is small and so the data provide little information about the group-level variance (Gelman, 2006).”
BackgroundUnconscious Visual Working Memory: A critical review and Bayesian meta-analysis
Gambarota, F., Tsuchiya, N., Pastore, M., Polito, N. D., & Sessa, P. (2022) · Neuroscience & Biobehavioral Reviews · doi:10.1016/j.neubiorev.2022.104618 · Cited by 23
Walker, T. J., Kohl, H. W., Bartholomew, J. B., Green, C., & Fernández, M. E. (2022) · Implementation Science Communications · doi:10.1186/s43058-022-00271-9 · Cited by 21
“Convergence of Bayesian analyses on the posterior distributions via Monte-Carlo Markov chain will be assessed via graphical (Trace Plot, Autocorrelation Plot) and quantitative evidence (Gelman-Rubin Diagnostics and Effective Sample Size Diagnostics).”
BackgroundInfluentialBeissinger, S. R., MacLean, S. A., Iknayan, K. J., & de Valpine, P. (2023) · Science Advances · doi:10.1126/sciadv.abn0250 · Cited by 25
Zhou, Y., Hu, J.-N., Hao, G., Zhu, Z.-R., & Zhang, J. (2023) · Journal of Bridge Engineering · doi:10.1061/jbenf2.beeng-6235 · Cited by 20
“To ensure the random effect of the error, the default noninformative prior distribution IGam (0.001, 0.001) is chosen for σ 2 ε (Gelman 2006; Klein and Kneib 2016).”
InfluentialSengupta, P. P., & Chakraborty, S. (2023) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2023.116448 · Cited by 18
Bayesian benchmark dose analysis
Fang, Q., Piegorsch, W. W., & Barnes, K. (2015) · Environmetrics · doi:10.1002/env.2339 · Cited by 22
Additive mixed models with Dirichlet process mixture and P-spline priors
Heinzl, F., Fahrmeir, L., & Kneib, T. (2011) · AStA Advances in Statistical Analysis · doi:10.1007/s10182-011-0161-6 · Cited by 18
Fujimoto, M., Nishihara, G. N., & Terada, R. (2014) · Fisheries Science · doi:10.1007/s12562-014-0750-x · Cited by 18
A pathway analysis method for genome‐wide association studies
Shahbaba, B., Shachaf, C. M., & Yu, Z. (2012) · Statistics in Medicine · doi:10.1002/sim.4477 · Cited by 18
Chen, J., Hubbard, S. S., Gaines, D., et al. (2010) · Water Resources Research · doi:10.1029/2009wr008715 · Cited by 18
Juntunen, T., Vanhatalo, J. P., Peltonen, H., et al. (2011) · ICES Journal of Marine Science · doi:10.1093/icesjms/fsr183 · Cited by 18
“We used half a Student’s t-distribution that works as a weakly informative prior (Gelman, 2006; Vanhatalo et al., 2010), assumed that the spatial correlations were a priori short, and adjusted the range of our beliefs with the scale of the Student’s t-distribution.”
Chang, J., Chen, Y., Halteman, W. A., & Wilson, C. J. (2015) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/cjfas-2015-0018 · Cited by 24
Improving Water Quality Assessments through a Hierarchical Bayesian Analysis of Variability
Gronewold, A. D., & Borsuk, M. E. (2010) · Environmental Science & Technology · doi:10.1021/es100657p · Cited by 18
No evidence for larger brains in cooperatively breeding cichlid fishes
Reddon, A. R., O’Connor, C. M., Ligocki, I. Y., Hellmann, J. K., Marsh‐Rollo, S. E., Hamilton, I. M., & Balshine, S. (2016) · Canadian Journal of Zoology · doi:10.1139/cjz-2015-0118 · Cited by 22
“This corresponds to an inverse-gamma distribution with shape and scale parameters equal to 0.01, which is canonical (Gelman 2006).”
BackgroundInfluentialBayesian multi-scale modeling for aggregated disease mapping data
Aregay, M., Lawson, A. B., Faes, C., et al. (2015) · Statistical Methods in Medical Research · doi:10.1177/0962280215607546 · Cited by 18
Extended convolution model to bayesian spatio-temporal for diagnosing the DHF endemic locations
Mukhsar, M., Abapihi, B., Sani, A., et al. (2016) · Journal of Interdisciplinary Mathematics · doi:10.1080/09720502.2015.1047591 · Cited by 18
Investigating trends in asthma and COPD through multiple data sources: A small area study
Boulieri, A., Hansell, A., & Blangiardo, M. (2016) · Spatial and Spatio-temporal Epidemiology · doi:10.1016/j.sste.2016.05.004 · Cited by 18
“A weakly informative half Normal prior N(0,1) is assigned to each of the parameters σh, σv and σγ (Gelman, 2006), while a N(α, β (2)) prior is assigned to the log(σ(2) ik) with parameters α and β(2) following priors N(0, 1000) and N(0, 2.”
BackgroundComparison of intra-day and inter-day variation in LIBS spectra
Gupta, A. K., Curran, J. M., Coulson, S. A., et al. (2017) · Forensic Chemistry · doi:10.1016/j.forc.2016.12.002 · Cited by 18
Clustering and variable selection in the presence of mixed variable types and missing data
Storlie, C. B., Myers, S. M., Katusic, S. K., Weaver, A. L., Voigt, R. G., Croarkin, P. E., Stoeckel, R. E., & Port, J. D. (2018) · Statistics in Medicine · doi:10.1002/sim.7697 · Cited by 22
Detection of non-thermal X-ray emission in the lobes and jets of Cygnus A
de Vries, M. N., Wise, M. W., Huppenkothen, D., et al. (2018) · Monthly Notices of the Royal Astronomical Society · doi:10.1093/mnras/sty1232 · Cited by 18
“This distribution includes the zero point, is heavy-tailed, and is recommended for scale parameters in Bayesian hierarchical models (Gelman 2006; Polson & Scott 2011).”
BackgroundKneib, T., Klein, N., Lang, S., & Umlauf, N. (2019) · Test · doi:10.1007/s11749-019-00631-z · Cited by 28
Chitakasempornkul, K., Meneget, M. B., Rosa, G. J. M., Lopes, F. B., Jager, A., Gonçalves, M., Dritz, S. S., Tokach, M. D., Goodband, R. D., & Bello, N. M. (2019) · Journal of Animal Science · doi:10.1093/jas/skz115 · Cited by 22
Fatigue-Related Changes in Running Gait Patterns Persist in the Days Following a Marathon Race
Clermont, C. A., Pohl, A. J., & Ferber, R. (2019) · Journal of Sport Rehabilitation · doi:10.1123/jsr.2019-0206 · Cited by 23
Henry, D. A. W., Lee, A. T., & Altwegg, R. (2020) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.13379 · Cited by 18
“The σx, σγ and σδ parameters were assigned a half-Cauchy prior (Gelman, 2006).”
Parameter clustering in Bayesian functional principal component analysis of neuroscientific data
Margaritella, N., Inácio, V., & King, R. (2020) · Statistics in Medicine · doi:10.1002/sim.8768 · Cited by 19
Evaluating post-vaccine expansion patterns of pneumococcal serotypes
Phillips, M. T., Warren, J. L., Givon‐Lavi, N., Tóthpál, A., Regev‐Yochay, G., Dagan, R., & Weinberger, D. M. (2020) · Vaccine · doi:10.1016/j.vaccine.2020.10.045 · Cited by 16
Turek, K., & Henkens, K. (2020) · Advances in Life Course Research · doi:10.1016/j.alcr.2020.100396 · Cited by 24
Prior Sample Size Extensions for Assessing Prior Impact and Prior-Likelihood Discordance
Reimherr, M. L., Meng, X., & Nicolae, D. L. (2021) · Journal of the Royal Statistical Society Series B (Statistical Methodology) · doi:10.1111/rssb.12414 · Cited by 42
“For more studies on weakly informative prior, see Gelman (2006) and Polson and Scott (2012).”
BackgroundMethodInfluentialBakhshi, A. K., & Ahmed, M. M. (2021) · IATSS Research · doi:10.1016/j.iatssr.2021.06.005 · Cited by 21
Grassetti, L., Bellio, R., Gaspero, L. D., Fonseca, G., & Vidoni, P. (2020) · IMA Journal of Management Mathematics · doi:10.1093/imaman/dpaa022 · Cited by 25
Bara, I., Darda, K. M., Kurz, A. S., & Ramsey, R. (2021) · European Journal of Neuroscience · doi:10.1111/ejn.15479 · Cited by 19
Bayesian Statistics Small Area Estimation
Gómez‐Rubio, V., Best, N., Richardson, S. T., Li, G., & Clarke, P. (2010) · Cited by 19
Ishigami, H. (2015) · Journal of Sports Sciences · doi:10.1080/02640414.2015.1039462 · Cited by 18
“…0; 100 2 ; σα,Uniformð0; 100Þ; and for βk is given by βk,Normal μβ; σ 2 β ; μβ,Normal 0; 100 2 ; σβ,half - Cauchyð5Þ: The priors for the variance parameters (σα 2 and σβ 2) are recommended by Gelman (2006), and other priors used in the analyses are also uninformative or weakly informative.”
Lada, H., Nally, R. M., & Taylor, A. C. (2008) · Journal of Mammalogy · doi:10.1644/07-mamm-a-080r.1 · Cited by 18
Technow, F., & Totir, L. R. (2015) · G3 Genes Genomes Genetics · doi:10.1534/g3.115.019299 · Cited by 20
“Thus, our results are consistent with the conclusion of Gelman (2006a), stating that the greatest potential of partial pooling with multilevel models is in predictive applications.”
BackgroundMethodResultInfluentialRubio, F. J., & Steel, M. F. J. (2013) · Bayesian Analysis · doi:10.1214/13-ba802 · Cited by 18
“This prior was proposed in Gelman (2006) as a weakly informative prior for this sort of parameters.”
BackgroundPaddock, S. M., & Louis, T. A. (2011) · Journal of the Royal Statistical Society Series C (Applied Statistics) · doi:10.1111/j.1467-9876.2010.00760.x · Cited by 18
“For the analysis, hyperparameters were set to a0 =a1 =b0 =b1 =0:0001; the results that were obtained were insensitive to this prior parameterization relative to other priors that we examined, including a0 =a1 =b0 =b1 =1 as well as specifying independent U.0, 1000/ priors for α and β (Gelman, 2006).”
MethodBayesian semi-parametric accelerated failure time model for paired doubly interval-censored data
Komárek, A., & Lesaffre, E. M. E. H. (2006) · Statistical Modelling · doi:10.1191/1471082x06st107oa · Cited by 18
“See Gelman (2006) for a general discussion on when the uniform prior is preferred to the inverse-gamma.”
BackgroundMcGuigan, K., Aguirre, J. D., & Blows, M. W. (2015) · Genetics · doi:10.1534/genetics.115.178632 · Cited by 19
“It is recommended that the scale hyperparameter be a value larger than what would be typically expected (Gelman 2006; Hadfield 2014); hence we specified h2M ¼ 0:01 based on the largest mutational heritability observed for morphologic traits in D. melanogaster (Houle et al. 1996; Lynch et al. 1999).”
BackgroundTransgenerational Adaptation to Pollution Changes Energy Allocation in Populations of Nematodes
Goussen, B., Péry, A. R. R., Bonzom, J., & Beaudouin, R. (2015) · Environmental Science & Technology · doi:10.1021/acs.est.5b03405 · Cited by 18
Lüdtke, O., Robitzsch, A., & Trautwein, U. (2018) · Multivariate Behavioral Research · doi:10.1080/00273171.2017.1406793 · Cited by 19
Modeling Chickenpox Dynamics with a Discrete Time Bayesian Stochastic Compartmental Model
Corberán‐Vallet, A., Santonja, F. J., Jornet, M., et al. (2018) · Complexity · doi:10.1155/2018/3060368 · Cited by 18
Energetically relevant predator–prey body mass ratios and their relationship with predator body size
Reum, J. C. P., Holsman, K. K., Aydin, K., Blanchard, J. L., & Jennings, S. (2018) · Ecology and Evolution · doi:10.1002/ece3.4715 · Cited by 18
“A half‐Cauchy prior was placed on the standard deviation of each random effect, with location and scale parameters set to 0 and 10, respectively (Gelman, 2006).”
MethodInfluentialThe Focal-Plane Detector Package on the TUNL Split-Pole Spectrograph
Marshall, C., Setoodehnia, K., Kowal, K., et al. (2018) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2018.2847938 · Cited by 18
“rized by , the location parameter, and , the scale parameter. The Half-Cauchy distribution has been found to give good behavior close to 0, and also avoids the hard limits of the Uniform distribution [38]. For this model ˝0˘HalfCauchy(0;1) was adopted, but results showed no noticeable dependence on . The comparison of…”
MethodGeneralised linear mixed model analysis via sequential Monte Carlo sampling
Fan, Y., Leslie, D. S., & WAND, M. P. (2008) · Research Online (University of Wollongong) · doi:10.48550/arxiv.0810.1163 · Cited by 18
Brown, D. A., Lazar, N. A., Datta, G. S., Jang, W., & McDowell, J. E. (2013) · NeuroImage · doi:10.1016/j.neuroimage.2013.08.024 · Cited by 21
Combining spatial information sources while accounting for systematic errors in proxiesPreprint
Paciorek, C. J. (2012) · Cited by 18
A Meta-Analysis of Relative Clause Processing in Mandarin Chinese Using Bias ModellingPreprint
Vasishth, S. (2024) · PsyArXiv (OSF Preprints) · doi:10.31234/osf.io/4un9k · Cited by 18
MCMC for inference on phase-type and masked system lifetime models
Aslett, L. J. M. (2012) · Arrow@dit (Dublin Institute of Technology) · Cited by 24
Towards stability of food production and farm income in a variable climate
Harkness, C., Areal, F., Semenov, M. A., Senapati, N., Shield, I. F., & Bishop, J. (2022) · Ecological Economics · doi:10.1016/j.ecolecon.2022.107676 · Cited by 36
Distributed Lag Interaction Models with Two Pollutants
Chen, Y., Mukherjee, B., & Berrocal, V. J. (2018) · Journal of the Royal Statistical Society Series C (Applied Statistics) · doi:10.1111/rssc.12297 · Cited by 21
“a0, b0/: We fix a0 = b0 = 0:001 to assume a non-informative hyperprior (Gelman, 2006).”
BackgroundMethodInfluentialHorseshoe prior Bayesian quantile regression
Kohns, D., & Szendrei, T. (2023) · Journal of the Royal Statistical Society Series C (Applied Statistics) · doi:10.1093/jrsssc/qlad091 · Cited by 21
“…artificially bias the error variances downwards when K >> T The general independent GL prior takes the following hierarchical form: As recommended by Gelman et al. (2006), we select a weakly informative inverse Gamma distribution as the prior for the error variance σ 2 and two independently…”
MethodBackfitting for large scale crossed random effects regressions
Ghosh, S., Hastie, T. J., & Owen, A. B. (2022) · The Annals of Statistics · doi:10.1214/21-aos2121 · Cited by 21
Ievoli, R., Gardini, A., & Palazzo, L. (2021) · AStA Advances in Statistical Analysis · doi:10.1007/s10182-021-00411-x · Cited by 17
“Eventually, in line with Gelman et al. (2006), half-Student’s t priors are set for the scale hyperparameters , and , choosing 3 degrees of freedom.”
BackgroundWang, Z., Lu, F., Liu, Z., Tu, W., Nie, K., Du, Q., Li, Q., & Wu, Z. (2022) · International Journal of Geographical Information Systems · doi:10.1080/13658816.2022.2117363 · Cited by 17
Exogeneous testosterone increases sexual impulsivity in heterosexual men
Wu, Y., Ou, J., Wang, X., Zilioli, S., Tobler, P. N., & Li, Y. (2022) · Psychoneuroendocrinology · doi:10.1016/j.psyneuen.2022.105914 · Cited by 20
Bayesian spatio-temporal analysis of the COVID-19 pandemic in Catalonia
Satorra, P., & Tebé, C. (2024) · Scientific Reports · doi:10.1038/s41598-024-53527-w · Cited by 21
Richardson, D. B., Laurier, D., Leuraud, K., Gillies, M., Haylock, R., Kelly‐Reif, K., Bertke, S. J., Daniels, R. D., Thierry-Chef, I., Moissonnier, M., et al. (2024) · American Journal of Epidemiology · doi:10.1093/aje/kwae256 · Cited by 20
Vo, T. D., Nishihara, G. N., Kitamura, Y., Shimada, S., Kawaguchi, S., & Terada, R. (2015) · Phycologia · doi:10.2216/14-61r1.1 · Cited by 17
“Weakly informative normal priors were placed on all of the parameters of the model, and a half-cauchy prior was placed on the scale parameter of the models (Gelman 2004, 2006).”
MethodBayesian analysis of multivariatetlinear mixed models with missing responses at random
Wang, W., & Lin, T. (2014) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2014.989852 · Cited by 17
Spatial analysis of bovine spongiform encephalopathy in Galicia, Spain (2000–2005)
Allepuz, A., López‐Quílez, A., Forte, A., et al. (2007) · Preventive Veterinary Medicine · doi:10.1016/j.prevetmed.2006.11.012 · Cited by 17
Hierarchical modeling of abundance in closed population capture–recapture models under heterogeneity
Schofield, M., & Barker, R. (2013) · Environmental and Ecological Statistics · doi:10.1007/s10651-013-0262-3 · Cited by 19
“The prior distributions for the standard deviations were specified using the redundant parameterization of Gelman (2006).”
MethodFarnham, D., & Lall, U. (2015) · Water Research · doi:10.1016/j.watres.2015.02.040 · Cited by 20
Individual level models vs. sample level models: contrasts and mutual benefits
Dumont, J., Giergiczny, M., & Hess, S. (2015) · Transportmetrica A Transport Science · doi:10.1080/23249935.2015.1018681 · Cited by 17
“This prior variance can have a large impact on the individual level posterior estimates, β (Gelman 2006).”
BackgroundPerformance of Sequential Imputation Method in Multilevel Applications
Zhao, E., Yucel, R. M., & Pearl, N. (2009) · Cited by 18
Using Combined Diagnostic Test Results to Hindcast Trends of Infection from Cross-Sectional Data
Rydevik, G., Innocent, G. T., Marion, G., Davidson, R. S., White, P. C. L., Billinis, C., Barrow, P., Mertens, P., Gavier‐Widén, D., & Hutchings, M. R. (2016) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1004901 · Cited by 16
King, R. (2011) · Chapman & Hall/CRC eBooks · doi:10.1201/b10905-18 · Cited by 21
Pöysä, H., Rintala, J. A., Johnson, D. H., et al. (2016) · Ecology and Evolution · doi:10.1002/ece3.2413 · Cited by 17
“…of the upper 50% of the normal distribution (values > 0); ki∼Uniform(kmin i, kmax i), where kmin i=Mean(ni)−2.576 ⋅SD(ni) and kmax i=Mean(ni)+2.576 ⋅SD(ni); Uniform (0, 10) was used for standard deviations τi and δi (for uniform noninformative priors, see Gelman, 2006; Kéry & Schaub, 2012).”
MethodInfluentialSpatiotemporal models for skewed processes
Schmidt, A. M., Gonçalves, K. C. M., & Velozo, P. L. (2016) · Environmetrics · doi:10.1002/env.2411 · Cited by 17
“For the variance of the measurement error, 𝜏, we suggest a half-Cauchy prior distribution as discussed by Gelman (2006); for the parameters in the exponential correlation function, 𝜙 and 𝜙𝜂 , we assign the same inverse gamma (IG) prior distribution, allowing the data to give evidence if there…”
MethodA comparative study of pricing approaches for longevity instruments
Leung, M., Fung, M. C., & O’Hare, C. (2018) · Insurance Mathematics and Economics · doi:10.1016/j.insmatheco.2018.06.010 · Cited by 17
Disturbance has benefits as well as costs for fragmented populations of a cryptic grassland reptile
Scroggie, M. P., Peterson, G. N. L., Rohr, D. H., et al. (2019) · Landscape Ecology · doi:10.1007/s10980-019-00865-0 · Cited by 17
“Vaguely informative Cauchy (0, 2.5) priors were specified for the regression parameters of the model (Gelman 2006; Gelman et al. 2008).”
MethodInfluentialAriyo, O., Lesaffre, E., Verbeke, G. N., & Quintero, A. (2019) · Communications in Statistics - Simulation and Computation · doi:10.1080/03610918.2019.1676439 · Cited by 17
“For instance, to obtain a minimally informative prior, k q appears appropriate (Gelman and Hill 2007; Gelman et al. 2014).”
BackgroundMethodInfluentialAljohani, H. M., & Alfaer, N. M. (2020) · Alexandria Engineering Journal · doi:10.1016/j.aej.2020.01.022 · Cited by 17
A Phase I–II Basket Trial Design to Optimize Dose-Schedule Regimes Based on Delayed Outcomes
Lin, R., Thall, P. F., & Yuan, Y. (2020) · Bayesian Analysis · doi:10.1214/20-ba1205 · Cited by 17
“To leverage information sharing, we follow the suggestion of Gelman (2006) by assuming a half-Cauchy ( ψ 0 , γ 0 ) prior on τ η in Level 3 of the model, where ψ 0 is the location parameter and γ 0 is the scale parameter.”
MethodMorales-Otero, M., & Núñez‐Antón, V. (2021) · Mathematics · doi:10.3390/math9030282 · Cited by 22
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.