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 1,101–1,200 of 4,089
Ito, T., Borlongan, I. A., Nishihara, G. N., Endo, H., & Terada, R. (2021) · Journal of Applied Phycology · doi:10.1007/s10811-021-02425-z · Cited by 25
“A half-cauchy prior was also placed on the scale parameter of the models (Gelman 2004, 2006).”
MethodOlivares, J., Sarro, L. M., Moraux, E., Berihuete, Á., Bouy, H., Hernández-Jiménez, S., Bertin, E., Galli, P. A. B., Huélamo, N., Bouvier, J., et al. (2018) · Astronomy and Astrophysics · doi:10.1051/0004-6361/201730972 · Cited by 25
Lam, V., Albrecht, M. A., Takechi, R., Giles, C., James, A. P., Foster, J. K., & Mamo, J. C. L. (2013) · Frontiers in Aging Neuroscience · doi:10.3389/fnagi.2013.00061 · Cited by 26
Tredennick, A. T., Bentley, L. P., & Hanan, N. P. (2013) · PLoS ONE · doi:10.1371/journal.pone.0058241 · Cited by 27
Estimation in the probit normal model for binary outcomes using the SAEM algorithm
Meza, C., Jaffrézic, F., & Foulley, J. (2008) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2008.11.024 · Cited by 16
Mark‐resight abundance estimation under incomplete identification of marked individuals
McClintock, B. T., Hill, J. M., Fritz, L. W., et al. (2013) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.12140 · Cited by 25
Common Dolphin (Delphinus delphis) Bycatch in New Zealand Commercial Trawl Fisheries
Thompson, F. N., Abraham, E. R., & Berkenbusch, K. (2013) · PLoS ONE · doi:10.1371/journal.pone.0064438 · Cited by 25
“Uninformative priors were given to all the parameters and hyperparameters (following recommendations from [36]).”
BackgroundTail Posterior Probability for Inference in Pairwise and Multiclass Gene Expression Data
Bochkina, N., & Richardson, S. T. (2007) · Biometrics · doi:10.1111/j.1541-0420.2007.00807.x · Cited by 26
“In this case, we adapt the recommendations of Gelman (2006) to use a uniform distribution on the standard deviation: 1/ √ b ∼ U [0, u].”
MethodThe ANTs Longitudinal Cortical Thickness PipelinePreprint
Tustison, N. J., Holbrook, A. J., Avants, B., Roberts, J. M., Cook, P. A., Reagh, Z. M., Duda, J., Stone, J. R., Gillen, D. L., Yassa, M. A., et al. (2017) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/170209 · Cited by 25
“where specification of variance priors to half-Cauchy distributions reflects commonly accepted best practice in the context of hierarchical models [72].”
MethodPenalising model component complexity: A principled, practical approach to constructing priorsPreprint
Simpson, D., Rue, H., Martins, T. G., et al. (2014) · arXiv (Cornell University) · doi:10.48550/arxiv.1403.4630 · Cited by 40
“Contrary to this, Gelman (2006), Gelman et al. (2008) argue that, when setting priors on the standard deviation of a normal distribution, heavy tails are necessary for robustness (O’Hagan and Pericchi, 2012). Unfortunately, this theory is only developed for inferring location-scale families or natural parameters in ex…”
BackgroundInfluentialShinohara, Y., Osaka, T., Inoue, I., et al. (n.d.) · RePEc: Research Papers in Economics · Cited by 25
Association between productivity and journal impact across disciplines and career age
Sunahara, A. S., Perc, M., & Ribeiro, H. V. (2021) · University of Maribor digital library (University of Maribor) · Cited by 25
Kiran, R. P., & Bansal, S. (2024) · Journal of Sound and Vibration · doi:10.1016/j.jsv.2024.118457 · Cited by 27
Ten questions concerning calibrating building energy simulation models
Chong, A., Yan, D., Sun, K., Zhan, S., Cheng, S., Wu, Y., Chen, Y., & Hong, T. (2025) · Building and Environment · doi:10.1016/j.buildenv.2025.113404 · Cited by 30
Evaluation of Bayesian spatiotemporal latent models in small area health data
Choi, J., Lawson, A. B., Cai, B., et al. (2011) · Environmetrics · doi:10.1002/env.1127 · Cited by 24
“The prior distributions of the intercept parameter and the variance parameters in the model are specified as ˛0 N.0; 2˛0/; ˛0 ; l ; w l ; p l Unif.0; d/ where 2˛0 is the variance and d is a constant (Gelman, 2006).”
BackgroundMoss, A. F. D., Juarez‐Colunga, E., Nathoo, F. S., Wagner, B. D., & Sagel, S. D. (2016) · Statistics in Medicine · doi:10.1002/sim.6845 · Cited by 26
Liu, F., & Li, Q. (2014) · Statistical Methods in Medical Research · doi:10.1177/0962280213519594 · Cited by 24
Baliga, V. B., & Mehta, R. S. (2016) · Integrative and Comparative Biology · doi:10.1093/icb/icw028 · Cited by 26
“For the prior, we used an inverse-Gamma distribution with shape and scale parameters equal to 0.01 (Gelman 2006).”
MethodInfluentialFrederiksen, M., Korner‐Nievergelt, F., Marion, L., et al. (2018) · Journal of Applied Ecology · doi:10.1111/1365-2664.13106 · Cited by 24
“For the standard deviation of the random year effects, we used folded- t distribution (M = 0, SD = 1, df = 2) priors (Gelman, 2006).”
MethodYu, J., & Baroud, H. (2019) · Risk Analysis · doi:10.1111/risa.13343 · Cited by 24
“The hyperpriors are generally specified by using expert elicitation, incorporating existing information, or using a proper noninformative distribution (Gelman, 2006).”
BackgroundAn Optimized Bayesian Hierarchical Two-Parameter Logistic Model for Small-Sample Item Calibration
König, C., Spoden, C., & Frey, A. (2019) · Applied Psychological Measurement · doi:10.1177/0146621619893786 · Cited by 28
Zhang, X., Gu, X., Wang, L., Zhou, Y., Huang, Z., Xu, C., & Cheng, C. (2020) · The Science of The Total Environment · doi:10.1016/j.scitotenv.2020.142626 · Cited by 24
“The term ε1i ~N (0, σε2) is the Gaussian random noise variable (Gelman, 2006).”
BackgroundWeber, F., Knapp, G., Glass, A. S., Kundt, G., & Ickstadt, K. (2020) · Research Synthesis Methods · doi:10.1002/jrsm.1471 · Cited by 26
“In the case of the “noninformative” prior for τ, we choose the U 0,∞ ð Þ distribution as it is the limiting case of the half-t distributions when the scale approaches infinity (and the degrees of freedom are fixed).(49) It is important to note that the U 0,∞ ð Þ prior only results in a proper posterior if K>2.”
Rosenberger, K. J., Xing, A., Murad, M. H., Chu, H., & Lin, L. (2021) · Journal of General Internal Medicine · doi:10.1007/s11606-020-06357-1 · Cited by 24
Bayesian model averaging for mortality forecasting using leave-future-out validation
Barigou, K., Goffard, P.-O., Loisel, S., & Salhi, Y. (2022) · International Journal of Forecasting · doi:10.1016/j.ijforecast.2022.01.011 · Cited by 32
“…x β x = 1 , we let the β x ’s be Dirichlet distributed with Since the model variance is measured by 1 /φ , see Equation (2.5), the parameter φ is actually a concentration parameter for which a standard prior assumption is the half-normal distribution, see for instance Gelman et al. (2006).”
BackgroundInfluentialZitzmann, S., Weirich, S., & Hecht, M. (2021) · Psych · doi:10.3390/psych3030025 · Cited by 26
Updating Generalizability Theory in Management Research: Bayesian Estimation of Variance Components
LoPilato, A. C., Carter, N. T., & Wang, M. (2014) · Journal of Management · doi:10.1177/0149206314554215 · Cited by 37
“Gelman (2006) provides a technical overview of the different prior distributions that can be used for a variance component, such 714 Journal of Management / February 2015 as the half-Cauchy distribution.”
BackgroundClimate Sensitivity Distributions Dependence on the Possibility that Models Share Biases
Lemoine, D. M. (2010) · Journal of Climate · doi:10.1175/2010jcli3503.1 · Cited by 23
Statistical models and computation to evaluate measurement uncertainty
Possolo, A. M. (2014) · Metrologia · doi:10.1088/0026-1394/51/4/s228 · Cited by 25
de la Cruz, R., & Branco, M. D. (2009) · Biometrical Journal · doi:10.1002/bimj.200800154 · Cited by 24
“Following the recommendations given by Gelman (2006), a uniform prior is considered for the scale parameter s, that is, s Uð0; asÞ.”
BackgroundMethodHow many hosts? Modelling host breadth from field samples
Vesk, P. A., Mccarthy, M. A., & Moir, M. L. (2010) · Methods in Ecology and Evolution · doi:10.1111/j.2041-210x.2010.00026.x · Cited by 24
Influence of district features on energy consumption in non-domestic buildings
Choudhary, R., & Tian, W. (2013) · Building Research & Information · doi:10.1080/09613218.2014.832559 · Cited by 24
“Following Gelman (2006), the prior on this variance parameter is specified as a uniform distribution between values 0 and 100.”
MethodBayesian estimation with integrated nested Laplace approximation for binary logit mixed models
Grilli, L., Metelli, S., & Rampichini, C. (2014) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2014.935377 · Cited by 24
“In the literature there is much debate on this issue.[11] An extensive comparison of different prior specifications is extremely time consuming with MCMC, but it is relatively fast with INLA.”
BackgroundMethodInfluentialJiao, Y., Rogers‐Bennett, L., Taniguchi, I. K., et al. (2010) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/f10-019 · Cited by 24
Bayesian Multilevel Model Calibration for Inverse Problems Under Uncertainty with Perfect Data
Nagel, J. B., & Sudret, B. (2015) · Journal of Aerospace Information Systems · doi:10.2514/1.i010264 · Cited by 31
“Adopting the latter paradigm, prior elicitation [18,19] and posterior computation [20,21] are delicate issues that have been discussed in the statistical literature.”
BackgroundReview and empirical comparison of joint mapping of multiple diseases
Manda, S., Feltbower, R. G., & Gilthorpe, M. S. (2012) · Southern African Journal of Epidemiology and Infection · doi:10.1080/10158782.2012.11441505 · Cited by 26
Spatial moving average risk smoothing
Botella-Rocamora, P., López‐Quílez, A., & Martínez‐Beneito, M. A. (2012) · Statistics in Medicine · doi:10.1002/sim.5704 · Cited by 24
EASY AND FLEXIBLE BAYESIAN INFERENCE OF QUANTITATIVE GENETIC PARAMETERS
Waldmann, P. (2009) · Evolution · doi:10.1111/j.1558-5646.2009.00645.x · Cited by 24
McKinnon, A. D., Doyle, J. R., Duggan, S., Logan, M., Lønborg, C., & Brinkman, R. (2015) · PLoS ONE · doi:10.1371/journal.pone.0140012 · Cited by 24
Dealing with Reflection Invariance in Bayesian Factor Analysis
Erosheva, E. A., & Curtis, S. M. (2017) · Psychometrika · doi:10.1007/s11336-017-9564-y · Cited by 36
“In addition, small estimates for factor variances may indicate that the resulting inference is sensitive to the choice of hyperparameters in the inverse-gamma prior distribution (Gelman, 2006; Gelman and Hill, 2007).”
BackgroundGregory, S. D., Nevoux, M., Riley, W. D., Beaumont, W. R. C., Jeannot, N., Lauridsen, R. B., Marchand, F., Scott, L. J., & Roussel, J. (2017) · Freshwater Biology · doi:10.1111/fwb.12929 · Cited by 28
Colclough, G. L., Woolrich, M. W., Harrison, S. J., Rojas-López, P. A., Valdes-Sosa, P. A., & Smith, S. M. (2018) · NeuroImage · doi:10.1016/j.neuroimage.2018.04.077 · Cited by 24
“Following advice in Gelman (2006) and Polson and Scott (2012) on the inference of higher-level group parameters in regression, we apply regularisation to the mean edge strengths (μ) towards zero using a normal distribution, learning the rough scale of these connectivities, χ2, from the data and…”
Fukumoto, R., Borlongan, I. A., Nishihara, G. N., et al. (2018) · Phycological Research · doi:10.1111/pre.12220 · Cited by 24
“Informative normal priors were placed on all of the parameters of the model by using values from previous studies (Kokubu et al. 2015), and a half-Cauchy prior was placed on the scale parameter of the models (Gelman 2004, 2006).”
MethodA Bayesian hidden Potts mixture model for analyzing lung cancer pathology images
Li, Q., Wang, X., Liang, F., Yi, F., Xie, Y., Gazdar, A. F., & Xiao, G. (2018) · Biostatistics · doi:10.1093/biostatistics/kxy019 · Cited by 23
Guul, T. S., Villadsen, A. R., & Wulff, J. N. (2019) · Public Administration Review · doi:10.1111/puar.13094 · Cited by 30
Time‐Varying Coefficient Models for the Analysis of Air Pollution and Health Outcome Data
Lee, D., & Shaddick, G. (2007) · Biometrics · doi:10.1111/j.1541-0420.2007.00776.x · Cited by 27
“Sensitivity analysis also shows that the posterior for τ 2 depends on the choice of Inverse-Gamma( , ) prior, and we use a noninformative prior on the standard deviation scale as suggested by Gelman (2006).”
MethodExploring the lipoprotein composition using Bayesian regression on serum lipidomic profiles
Sysi‐Aho, M., Vehtari, A., Velagapudi, V., Westerbacka, J., Yetukuri, L., Bergholm, R., Taskinen, M., Yki‐Järvinen, H., & Orešič, M. (2007) · Bioinformatics · doi:10.1093/bioinformatics/btm181 · Cited by 24
“For shg, sag and seg, the scale parameters of the t-distributions, we set half-Cauchy hyperpriors as suggested by Gelman (2006). For vhg, vag and veg we set power-law hyperpriors 1/ ( 42).”
MethodInfluentialA solution to separation for clustered binary data
Abrahantes, J. C., & Aerts, M. (2012) · Statistical Modelling · doi:10.1177/1471082x1001200102 · Cited by 24
“Gelman (2006) states that if the variance of the random effect is estimated to be near zero, the resulting inferences will be sensitive to �.”
BackgroundMethodA Cross-Cohort Changepoint Model for Customer-Base Analysis
Gopalakrishnan, A., Bradlow, E. T., & Fader, P. S. (2016) · Marketing Science · doi:10.1287/mksc.2016.1007 · Cited by 25
“Each mean‐variance pair is assumed to be Normal‐Inverse‐Chisquared (NIX) distributed for conjugacy, with relatively non‐informative yet proper hyperpriors (Gelman 2006).”
Obesity rise plateaus in developed nations and accelerates in developing nations
Zhou, B., Phelps, N. H., Galeazzi, A., O'Driscoll, O. N., Bennett, J. E., Jain, L., Brisis, Y. D. D., Barradas‐Pires, A., Deo, F., Stevens, G. A., et al. (2026) · Nature · doi:10.1038/s41586-026-10383-0 · Cited by 68
“The κ terms were each assigned a flat prior on the standard deviation scale 84 .”
Lauderdale, B. E., & Linzer, D. A. (2015) · International Journal of Forecasting · doi:10.1016/j.ijforecast.2015.03.002 · Cited by 23
“States’ party affinities drift over time, so we allow the magnitude of α st to vary from one election to the next as a Bayesian dynamic linear model (West & Harrison, 1997): ( is determined by an estimated σ α , over which we place a half-normal prior (Gelman, 2006).”
BackgroundOrme, M. E., Fenici, P., Duprat-Lomon, I., Wygant, G. D., Townsend, R., & Roudaut, M. (2014) · Diabetology & Metabolic Syndrome · doi:10.1186/1758-5996-6-73 · Cited by 26
Schmidli, H., Neuenschwander, B. E., & Friede, T. (2016) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2016.08.007 · Cited by 27
Multiple Principals and Legislative Cohesion
Meserve, S. A., Robbins, J. W., & Thames, F. C. (2017) · Legislative Studies Quarterly · doi:10.1111/lsq.12165 · Cited by 24
Addressing the estimation of standard errors in fixed effects meta‐analysis
Islas, C. P. D., & Rice, K. M. (2018) · Statistics in Medicine · doi:10.1002/sim.7625 · Cited by 25
Scalable inference for crossed random effects models
Papaspiliopoulos, O., Roberts, G. O., & Zanella, G. (2019) · Biometrika · doi:10.1093/biomet/asz058 · Cited by 28
Anjoy, P., Chandra, H., & Basak, P. (2018) · Social Indicators Research · doi:10.1007/s11205-018-2050-9 · Cited by 23
Achieving parsimony in Bayesian vector autoregressions with the horseshoe prior
Follett, L., & Yu, C. L. (2019) · Econometrics and Statistics · doi:10.1016/j.ecosta.2018.12.004 · Cited by 23
Mauritzen, J. (2020) · Energy Policy · doi:10.1016/j.enpol.2020.111489 · Cited by 30
“Use of the Cauchy prior distribution also allows for inference in the case of complete separation by covariates (Gelman, 2006). µα ∼ Cauchy(0, 2.5) (5) µβ 0 ∼ Cauchy(0, 2.5) σα ∼ half-Cauchy(0, 5) σβ 0 ∼ half-Cauchy(0, 5) 10The Cauchy distribution is a t-distribution with 1 degree of freedom.”
BackgroundMethodInfluentialThe impact of covariance priors on arm‐based Bayesian network meta‐analyses with binary outcomes
Wang, Z., Lin, L., Hodges, J. S., & Chu, H. (2020) · Statistics in Medicine · doi:10.1002/sim.8580 · Cited by 25
Virgillito, C., Manica, M., Marini, G., Rosà, R., della Torre, A., Martini, S., Drago, A., Baseggio, A., & Caputo, B. (2022) · Journal of the American Mosquito Control Association · doi:10.2987/21-7024 · Cited by 23
“Diffuse normal priors were used for regression parameters, while Cauchy priors were used for standard deviation parameters (Gelman 2006).”
MethodLee, S. Y. (2022) · Mathematics · doi:10.3390/math10060898 · Cited by 31
Li, C., Ferro, A., Mhatre, S. K., Lu, D., Lawrance, M., Li, X., Li, S., Allen, S. M., Desai, J., Fakih, M., et al. (2022) · Communications Medicine · doi:10.1038/s43856-022-00155-y · Cited by 27
Low-dimensional encoding of decisions in parietal cortex reflects long-term training history
Latimer, K. W., & Freedman, D. J. (2023) · Nature Communications · doi:10.1038/s41467-023-36554-5 · Cited by 30
Joint Modelling of Survival and Cognitive Decline in the Australian Longitudinal Study of Ageing
Graham, P. L., RYAN, L. M., & Luszcz, M. A. (2010) · Journal of the Royal Statistical Society Series C (Applied Statistics) · doi:10.1111/j.1467-9876.2010.00737.x · Cited by 23
“…analysis was undertaken in which (a) interactions and time varying covariates were examined, (b) different distributional specifications were placed on the survival model and (c) an alternative specification of R was used and uniform priors were placed on the variance components (Gelman, 2006).”
MethodQuick, H. S., Holan, S. H., Wikle, C. K., et al. (2015) · Spatial Statistics · doi:10.1016/j.spasta.2015.07.008 · Cited by 23
WHAT DO WE KNOW ABOUT THE HERITABILITY OF DEVELOPMENTAL INSTABILITY? ANSWERS FROM A BAYESIAN MODEL
Dongen, S. V. (2007) · Evolution · doi:10.1111/j.1558-5646.2007.00096.x · Cited by 23
Haining, R., Li, G., Maheswaran, R., Blangiardo, M., Law, J., Best, N., & Richardson, S. T. (2010) · Spatial and Spatio-temporal Epidemiology · doi:10.1016/j.sste.2010.03.006 · Cited by 22
“We have assessed the influence of the prior specification for the random effect variance by using two vague priors, a) a Gamma(0.5,0.005) on the precision and b) a uniform(0.01,1) on the standard deviation (Gelman 2006).”
MethodInfluentialInferring lung cancer risk factor patterns through joint Bayesian spatio-temporal analysis
Cramb, S. M., Baade, P. D., White, N. M., RYAN, L. M., & Mengersen, K. L. (2015) · Cancer Epidemiology · doi:10.1016/j.canep.2015.03.001 · Cited by 27
A Bayesian analysis of the Gumbel distribution: an application to extreme rainfall data
Vidal, I. (2013) · Stochastic Environmental Research and Risk Assessment · doi:10.1007/s00477-013-0773-3 · Cited by 23
“Other prior distributions could also be considered, for example the Scaled Beta2 distribution [see Perez and Pericchi (2012) and Fuquene et al. (2013)] or the half-t (see Gelman (2006)), among other.”
BackgroundHallander, J., Waldmann, P., Wang, C., & Sillanpää, M. J. (2010) · Genetics · doi:10.1534/genetics.110.114249 · Cited by 26
“Bayesian inference of the factorized model: Following Waldmann (2009), uniform distributions were assigned as priors to the standard deviations, since this prior distribution was shown to be a good and robust noninformative choice of prior for variance components (Gelman 2006).”
MethodChoi, J., Lawson, A. B., Cai, B., et al. (2012) · Statistical Methods in Medical Research · doi:10.1177/0962280212446318 · Cited by 23
Lockwood, J. R., Savitsky, T. D., & McCaffrey, D. F. (2015) · The Annals of Applied Statistics · doi:10.1214/15-aoas833 · Cited by 32
“We also tested sensitivity to an alternative prior distribution where the square roots of the udd were modeled as IID uniform [Gelman (2006)].”
MethodPenalized complexity priors for degrees of freedom in Bayesian P-splines
Ventrucci, M., & Rue, H. (2016) · Statistical Modelling · doi:10.1177/1471082x16659154 · Cited by 24
“The suitability of the gamma family as a non-informative prior for the scale parameters in hierarchical models has been debated in the literature (Gelman, 2006); overfitting due to gamma priors has been demonstrated in Frühwirth-Schnatter and Wagner (2010, 2011) and Simpson et al.”
MethodPlant–soil feedbacks and the dominance of conifers in a tropical montane forest in Borneo
Ushio, M., Aiba, S., Takeuchi, Y., Iida, Y., Matsuoka, S., Repin, R., & Kitayama, K. (2016) · Ecological Monographs · doi:10.1002/ecm.1236 · Cited by 24
“Half- cauchy prior distributions were used for variances of random effects in this model (Gelman 2006).”
MethodLow-cost energy meter calibration method for measurement and verification
Carstens, H., Xia, X., & Yadavalli, V. S. S. (2016) · Applied Energy · doi:10.1016/j.apenergy.2016.12.028 · Cited by 26
Webb, J. A., de Little, S. C., Miller, K. A., & Stewardson, M. J. (2018) · Freshwater Biology · doi:10.1111/fwb.13069 · Cited by 25
“For these parameters, the standard deviation among 214 sites within rivers, and among rivers were both fitted at the state level; there were insufficient 215 replicate sites within rivers to reliably estimate between-site variation within each river separately 216 (Gelman, 2006).”
BackgroundBayesian evaluation of behavior change interventions: a brief introduction and a practical example
Heino, M. T. J., Vuorre, M., & Hankonen, N. E. (2018) · Health Psychology and Behavioral Medicine · doi:10.1080/21642850.2018.1428102 · Cited by 30
“This distribution is a positive only Cauchy distribution with scale 1 (Gelman, 2006).”
Using Bayesian Latent Gaussian Graphical Models to Infer Symptom Associations in Verbal Autopsies
Li, Z. R., McCormick, T. H., & Clark, S. J. (2019) · Bayesian Analysis · doi:10.1214/19-ba1172 · Cited by 21
“In this paper we adopt the improper prior on the marginal standard deviations suggested in Gelman (2006), so that Λ jj ∝ 1.”
MethodMathieu‐Bégné, E., Loot, G., Chevalier, M., Paz‐Vinas, I., & Blanchet, S. (2018) · Oikos · doi:10.1111/oik.05511 · Cited by 24
Confronting Quasi-Separation in Logistic Mixed Effects for Linguistic Data: A Bayesian Approach
Kimball, A. E., Shantz, K., Eager, C. D., & Roy, J. (2018) · Journal of Quantitative Linguistics · doi:10.1080/09296174.2018.1499457 · Cited by 34
“The median of a half Cauchy distribution is the scale parameter itself, and 75% of the probability density is between 0 and 2.5 times the scale (Gelman, 2006).”
BackgroundInfluentialTerada, R., Nakashima, Y., Borlongan, I. A., Shimabukuro, H., Kozono, J., Endo, H., Shimada, S., & Nishihara, G. N. (2019) · Phycological Research · doi:10.1111/pre.12398 · Cited by 23
“Informative normal priors were placed on all of the parameters of the model by using values from previous studies (Kokubu et al. 2015), and a half-cauchy prior was placed on the scale parameter of the models (Gelman 2004, 2006).”
MethodBayesian regularization for flexible baseline hazard functions in Cox survival models
Lázaro, E., Armero, C., & Alvares, D. (2020) · Biometrical Journal · doi:10.1002/bimj.201900211 · Cited by 27
“As mentioned previously, a usual choice for the hyperprior distribution for σ 2 γ is an inverse gamma distribution or also a proper uniform distribution (Gelman, 2006).”
BackgroundMethodA Bayesian dose-response meta-analysis model: A simulations study and application.
Hamza, T., Cipriani, A., Furukawa, T. A., Egger, M., Orsini, N., & Salanti, G. (2021) · doi:10.48350/151952 · Cited by 23
Objective Bayesian Analysis of a Measurement Error Small Area Model
Arima, S., Datta, G. S., & Liseo, B. (2012) · Bayesian Analysis · doi:10.1214/12-ba712 · Cited by 23
MacTavish, R., Bixby, H., Cavanaugh, A., Agyei-Mensah, S., Bawah, A., Owusu, G., Ezzati, M., Arku, R., Robinson, B., Schmidt, A. M., et al. (2023) · World Development · doi:10.1016/j.worlddev.2023.106253 · Cited by 23
“The prior for the Vj follows independent zero mean normal distributions, with a weakly-informative half-Cauchy prior for the standard deviation (Gelman, 2006).”
Michiels, B., Tanious, R., De, T. K., & Onghena, P. (2020) · Behavior Research Methods · doi:10.3758/s13428-019-01266-6 · Cited by 23
AI contextual information shapes moral and aesthetic judgments of AI-generated visual art
Bara, I., Ramsey, R., & Cross, E. S. (2025) · Cognition · doi:10.1016/j.cognition.2025.106063 · Cited by 37
Park, S., & Kaplan, D. E. (2015) · Multivariate Behavioral Research · doi:10.1080/00273171.2014.1003770 · Cited by 27
“Gelman (2006) noted that serious problems can occur with the inverse-gamma family of non-informative prior distributions, and recommends to use the uniform prior when the number of groups is small and in other settings where weakly informative priors are necessary.”
BackgroundMethodWang, Z., Du, Q., Shi, L., Nie, K., Lin, D., Chen, Y., & Li, J. (2014) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph110100713 · Cited by 22
Trouvé, R., Nitschke, C. R., Andrieux, L., et al. (2019) · Forest Ecology and Management · doi:10.1016/j.foreco.2019.05.055 · Cited by 22
“We used weakly informative priors (Gelman, 2006) for all models.”
MethodBayesian Estimates of Transmission Line Outage Rates That Consider Line Dependencies
Zhou, K., Cruise, J. R., Dent, C. J., Dobson, I., Wehenkel, L. A., Wang, Z., & Wilson, A. L. (2020) · IEEE Transactions on Power Systems · doi:10.1109/tpwrs.2020.3012840 · Cited by 27
“Gelman [14], however, does not recommend the inversegamma prior as the estimation of σ(2) would be sensitive to the parameters of inverse-gamma distribution when σ(2) is near zero.”
BackgroundMortality Models Incorporating Long Memory Improves Life Table Estimation: A Comprehensive Analysis
Yan, H., Peters, G. W., & Chan, J. (2018) · Annals of Actuarial Science · doi:10.1017/s1748499521000014 · Cited by 26
Interrater reliability for multilevel data: A generalizability theory approach.
ten Hove, D., Jorgensen, T. D., & van der Ark, L. A. (2022) · Psychological Methods · doi:10.1037/met0000391 · Cited by 29
van Gils, M. J., van Willigen, H. D. G., Wynberg, E., Han, A. X., van der Straten, K., Burger, J. A., Poniman, M., Oomen, M., Tejjani, K., Bouhuijs, J. H., et al. (2021) · Cell Reports Medicine · doi:10.1016/j.xcrm.2021.100486 · Cited by 32
The epidemiology of gonorrhoea in London: a Bayesian spatial modelling approach
de Waroux, O. L. P., Harris, R., Hughes, G., & Crook, P. (2013) · Epidemiology and Infection · doi:10.1017/s0950268813000745 · Cited by 25
“Medians and 95% credible intervals (CrIs) were obtained on the basis of 20000 iterations after a 5000 iteration ‘burn-in’ period, with convergence assessed via the Gelman–Rubin diagnostic [39].”
MethodBayesian variable selection for probit mixed models applied to gene selection
Baragatti, M. (2011) · Bayesian Analysis · doi:10.1214/11-ba607 · Cited by 23
“For the prior distributions for σ2 1 , we chose a IG(2, 3) which seemed reasonable given our data, a IG(2, 5) to have a prior favoring higher values compared to the first one, a IG(3, 1) to favor lower values, and a IG(1, 1) to have a non-informative prior without too small parameters to avoid problems, see Gelman [20…”
BackgroundMethodMultidimensional Assessment of Value Added by Teachers to Real-World Outcomes
Broatch, J., & Lohr, S. L. (2011) · Journal of Educational and Behavioral Statistics · doi:10.3102/1076998610396900 · Cited by 22
“The estimates of G and R, however, may be sensitive to the choice of prior distribution (Gelman, 2006).”
BackgroundHabitat connectivity, complexity, and quality: predicting adult coho salmon occupancy and abundance
Anlauf‐Dunn, K. J., Ward, E. J., Strickland, M., & Jones, K. K. (2014) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/cjfas-2014-0162 · Cited by 27
Spatial and Temporal Variation in the Effects of Climatic Variables on Dugong Calf Production
Fuentes, M. M. P. B., Delean, S., Grayson, J., Lavender, S. L., Logan, M., & Marsh, H. (2016) · PLoS ONE · doi:10.1371/journal.pone.0155675 · Cited by 26
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.