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 601–700 of 4,089
Gundale, M. J., Fajardo, A., Lucas, R. W., Nilsson, M., & Wardle, D. A. (2011) · Ecography · doi:10.1111/j.1600-0587.2011.06853.x · Cited by 47
“We assigned non-informative, uniform priors to all unknown parameters and used folded-Cauchy densities (i.e. a Student’s t-distribution with one degree of freedom) as priors for all standard deviations (s%1/!t, where t is the precision parameter of interest) (Gelman 2006).”
MethodGsteiger, S., Neuenschwander, B. E., Mercier, F., & Schmidli, H. (2013) · Statistics in Medicine · doi:10.1002/sim.5851 · Cited by 52
Integrating occupancy models and structural equation models to understand species occurrence
Joseph, M. B., Preston, D. L., & Johnson, P. T. J. (2015) · Ecology · doi:10.1890/15-0833.1 · Cited by 53
Estimation of Risk Ratios in Cohort Studies With Common Outcomes
Chu, H., & Cole, S. R. (2010) · Epidemiology · doi:10.1097/ede.0b013e3181f2012b · Cited by 47
Terada, R., Matsumoto, K., Borlongan, I. A., Watanabe, Y., Nishihara, G. N., Endo, H., & Shimada, S. (2017) · Journal of Applied Phycology · doi:10.1007/s10811-017-1344-7 · Cited by 47
Deng, H., Devleesschauwer, B., Liu, M., Li, J., Wu, Y., van der Giessen, J. W. B., & Opsteegh, M. (2018) · Scientific Reports · doi:10.1038/s41598-018-24361-8 · Cited by 56
Melanson, J. E., Thibeault, M.-P., Stocks, B. B., Leek, D. M., McRae, G., & Meija, J. (2018) · Analytical and Bioanalytical Chemistry · doi:10.1007/s00216-018-1272-7 · Cited by 48
Li, L., Zhang, J., Meng, X., Fang, Y., Ge, Y., Wang, J., Wang, C., Wu, J. Y., & Kan, H. (2018) · Remote Sensing of Environment · doi:10.1016/j.rse.2018.09.001 · Cited by 48
Penjor, U., Wangdi, S., Tandin, T., & Macdonald, D. W. (2020) · Ecological Indicators · doi:10.1016/j.ecolind.2020.107085 · Cited by 51
Controls of outbursts of moraine-dammed lakes in the greater Himalayan region
Fischer, M., Korup, O., Veh, G., & Walz, A. (2021) · The Cryosphere · doi:10.5194/tc-15-4145-2021 · Cited by 50
Effectiveness of fish waste management strategies in reducing seabird attendance at a trawl vessel
Abraham, E. R., Pierre, J. P., Middleton, D. A. J., et al. (2008) · Fisheries Research · doi:10.1016/j.fishres.2008.08.014 · Cited by 47
A Bayesian Multivariate Functional Dynamic Linear Model
Kowal, D. R., Matteson, D. S., & Ruppert, D. (2016) · Journal of the American Statistical Association · doi:10.1080/01621459.2016.1165104 · Cited by 67
“This is similar to the hierarchical setting of Gelman (2006), in which there are M + 2 groups for each λk, k = 1, . . . ,K.”
BackgroundMethodInfluentialTan, A., & Hobert, J. P. (2009) · Journal of Computational and Graphical Statistics · doi:10.1198/jcgs.2009.08153 · Cited by 44
“In contrast, Gelman (2006) illustrated that the improper prior (σ 2θ ) −1/2 works well unless q is very small (say, below 5).”
BackgroundInfluentialIntegrating the niche and neutral perspectives on community structure and dynamics
Mutshinda, C. M., & O’Hara, R. B. (2010) · Oecologia · doi:10.1007/s00442-010-1831-x · Cited by 45
“Gelman (2006) gives a justification for this approach to specifying priors for variance parameters.”
BackgroundWeak Informativity and the Information in One Prior Relative to Another
Evans, M. J., & Jang, G. H. (2011) · Statistical Science · doi:10.1214/11-sts357 · Cited by 69
Analysis of the spatial variation of Bovine tuberculosis disease risk in Spain (2006–2009)
Allepuz, A., Casal, J., Napp, S., Sáez, M., Alba, A., Vilar, M. J., Domingo, M., González, M. Á., Durán‐Ferrer, M., Vicente, J., et al. (2011) · Preventive Veterinary Medicine · doi:10.1016/j.prevetmed.2011.02.012 · Cited by 46
Stewart, K. R., Sims, M., Meylan, A. B., Witherington, B. E., Brost, B., & Crowder, L. B. (2010) · Ecological Applications · doi:10.1890/09-1838.1 · Cited by 56
Epigenetic Control of Phenotypic Plasticity in the Filamentous Fungus Neurospora crassa
Kronholm, I., Johannesson, H., & Ketola, T. (2016) · G3 Genes Genomes Genetics · doi:10.1534/g3.116.033860 · Cited by 51
“The model speci fi cation followed Gelman (2006) and Kruschke (2011): where b 0 is the grand mean, b j is the effect of the j th genotype G .”
BackgroundMethodNot Normal: the uncertainties of scientific measurements
Bailey, D. (2017) · Royal Society Open Science · doi:10.1098/rsos.160600 · Cited by 55
Challenging Conventional Wisdom for Multivariate Statistical Models With Small Samples
McNeish, D. (2017) · Review of Educational Research · doi:10.3102/0034654317727727 · Cited by 57
“…to help inform prior distributions, though some may view this strategy as inappropriate because it “double-dips” the data (see, e.g., McNeish, 2016c). gelman (2006) also advanced a half-Cauchy prior that has shown promise as a fairly noninformative prior with small samples, although it can be…”
BackgroundJardine, E. (2017) · New Media & Society · doi:10.1177/1461444817733134 · Cited by 46
“Yet others, such as Nezlek (2008) and Gelman (2006), maintain that fewer than 10 secondlevel units can be sufficient to justify the use of hierarchical models, even if there may be attendant estimation and interpretation problems for the second-level units.”
BackgroundThe biomechanical origin of extreme wing allometry in hummingbirds
Skandalis, D. A., Segre, P. S., Bahlman, J. W., Groom, D. J. E., Welch, K. C., Witt, C. C., McGuire, J. A., Dudley, R., Lentink, D., & Altshuler, D. L. (2017) · Nature Communications · doi:10.1038/s41467-017-01223-x · Cited by 51
An optimal Bayesian regularization for force reconstruction problems
Aucejo, M., & de Smet, O. (2019) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2019.02.021 · Cited by 46
“According to Gelman, a prior distribution is said weakly informative if it is proper but is set up so that the information it does provide is intentionally weaker than whatever actual prior knowledge is available [43].”
BackgroundSmid, S. C., & Winter, S. D. (2020) · Frontiers in Psychology · doi:10.3389/fpsyg.2020.611963 · Cited by 60
“For small sample sizes, the combination of the relatively larger impact of the prior on the posterior and the wide distribution of default priors can lead to extremely incorrect parameter estimates (see e.g., Gelman, 2006; McNeish, 2016a; and the systematic literature review of Smid et al., 2019b).”
BackgroundMethodJaya, I. G. N. M., & Folmer, H. (2021) · Journal of Regional Science · doi:10.1111/jors.12533 · Cited by 46
Saber, M., Giordano, K. R., Hur, Y., Ortiz, J. B., Morrison, H., Godbout, J. P., Murphy, S. M., Lifshitz, J., & Rowe, R. K. (2020) · European Journal of Neuroscience · doi:10.1111/ejn.14611 · Cited by 46
“Following the recommendations by Gelman (2006) for priors on variance components, we applied a half-Cauchy prior, ~Cauchy(0, 5), to the residual variance parameter (σ) of the log-normal model.”
MethodModeling Dynamic Preferences: A Bayesian Robust Dynamic Latent Ordered Probit Model
Stegmueller, D. (2013) · Political Analysis · doi:10.1093/pan/mpt001 · Cited by 52
“Gelman (2006) recommends this prior over the more commonly used inverse Gamma specification (Spiegelhalter et al. 1997): ffiffiffiffiffi 2 q Uð0, 10Þ: ð10Þ 8 It facilitates a simple specification test of the appropriateness of assuming independence of initial conditions and unobserved individual…”
MethodHeuck, C., Herrmann, C., Schabo, D. G., Brandl, R., & Albrecht, J. (2016) · Ibis · doi:10.1111/ibi.12444 · Cited by 50
Taboada, F. G., Stock, C. A. A., Griffies, S. M., Dunne, J. P. P., John, J. G., Small, J., & Tsujino, H. (2018) · Ocean Modelling · doi:10.1016/j.ocemod.2018.11.003 · Cited by 47
Swinton, P. A., Burgess, K., Hall, A., Greig, L., Psyllas, J., Aspe, R., Maughan, P., & Murphy, A. (2022) · Journal of Sports Sciences · doi:10.1080/02640414.2022.2128548 · Cited by 45
Socioeconomic inequalities in injury mortality in small areas of 15 European cities
Gotsens, M., Marí-Dell’Olmo, M., Pérez, K., Palència, L., Martínez‐Beneito, M. A., Rodríguez‐Sanz, M., Burström, B., Costa, G., Deboosere, P., Domínguez‐Berjón, M. F., et al. (2013) · Health & Place · doi:10.1016/j.healthplace.2013.09.003 · Cited by 50
Visual Categorization of Natural Movies by Rats
Vinken, K., Vermaercke, B., & de Beeck, H. O. (2014) · Journal of Neuroscience · doi:10.1523/jneurosci.3663-13.2014 · Cited by 54
“Uniform priors ranging from 0 to 100 were chosen for the SDs that were estimated in the model (as in Gelman, 2006).”
Lawson, A. B., Choi, J., Cai, B., Hossain, M., Kirby, R. S., & Liu, J. (2012) · Journal of Agricultural Biological and Environmental Statistics · doi:10.1007/s13253-012-0100-3 · Cited by 44
“Gelman (2004) presented the computational and practical benefits for this plug-in approach in comparison to a joint model.”
Ruiz, J., Prieto, L., & Astorga, D. (2012) · Ecological Modelling · doi:10.1016/j.ecolmodel.2012.03.019 · Cited by 50
Reduced availability of rhizobia limits the performance but not invasiveness of introduced A cacia
Wandrag, E. M., SHEPPARD, A. W., Duncan, R. P., et al. (2013) · Journal of Ecology · doi:10.1111/1365-2745.12126 · Cited by 45
“We assigned the overall intercept and regression coefficients normal prior distributions with mean 0 and variance 1000, and for the ‘site’ term, we specified a non-informative uniform prior (0–100) on the standard deviation following Gelman (2006).”
MethodBayesian Nonparametric Weighted Sampling Inference
Si, Y., Pillai, N. S., & Gelman, A. (2015) · Bayesian Analysis · doi:10.1214/14-ba924 · Cited by 87
Self-citationRoss, C. T., Strimling, P., Ericksen, K. P., Lindenfors, P., & Mulder, M. B. (2016) · Human Nature · doi:10.1007/s12110-015-9244-5 · Cited by 56
A partially non-ergodic ground-motion prediction equation for Europe and the Middle East
Kuehn, N. M., & Scherbaum, F. (2016) · Bulletin of Earthquake Engineering · doi:10.1007/s10518-016-9911-x · Cited by 49
Gorzo, J. M., Pidgeon, A. M., Thogmartin, W. E., Allstadt, A. J., Radeloff, V. C., Heglund, P. J., & Vavrus, S. J. (2016) · Ornithological Applications · doi:10.1650/condor-15-180.1 · Cited by 45
“We modeled nuisance effects such as year, route, and observer as independent random variables and assigned standard noninformative priors via the half-Cauchy distribution to the hyperparameters (Gelman 2006).”
MethodAcceptance of novel food by horses: The influence of food cues and nutrient composition
van den Berg, M., Giagos, V., Lee, C., Brown, W. Y., & Hinch, G. N. (2016) · Applied Animal Behaviour Science · doi:10.1016/j.applanim.2016.07.005 · Cited by 51
“mean 0 and variance which in turn followed a Half-Cauchy prior (Gelman, 2006).”
BackgroundThe Coach-Athlete Relationship and Athlete Psychological Outcomes
McGee, V., & DeFreese, J. D. (2018) · Journal of Clinical Sport Psychology · doi:10.1123/jcsp.2018-0010 · Cited by 71
Reproduction of East-African bats may guide risk mitigation for coronavirus spillover
Montecino‐Latorre, D., Goldstein, T., Gilardi, K. V. K., Wolking, D. J., VanWormer, E., Kazwala, R. R., Ssebide, B., Nziza, J., Sijali, Z., Cranfield, M. R., et al. (2020) · One Health Outlook · doi:10.1186/s42522-019-0008-8 · Cited by 45
Wonky whales: the evolution of cranial asymmetry in cetaceans
Coombs, E. J., Clavel, J., Park, T., Churchill, M., & Goswami, A. (2020) · BMC Biology · doi:10.1186/s12915-020-00805-4 · Cited by 50
Data augmentation for non-Gaussian regression models using variance-mean mixtures
Polson, N., & Scott, J. G. (2013) · Biometrika · doi:10.1093/biomet/ass081 · Cited by 45
“These latter steps exploit standard results on variance components in linear models, which can be found in, for example, Gelman (2006).”
BackgroundA Bayesian hierarchical spatio-temporal model for significant wave height in the North Atlantic
Vanem, E., Huseby, A. B., & Natvig, B. (2011) · Stochastic Environmental Research and Risk Assessment · doi:10.1007/s00477-011-0522-4 · Cited by 47
“The variance of the noise terms are all assigned inverse gamma3 conjugate priors, i.e. r2i IGðai; biÞ for i ¼ Z; l; h;m and T: It is acknowledged that Gelman (2006) recommended uniform rather than inverse gamma priors for the variances to avoid constraints on the posteriors, but inverse gamma…”
BackgroundNewly Hatched Chicks Solve the Visual Binding Problem
Wood, J. N. (2014) · Psychological Science · doi:10.1177/0956797614528955 · Cited by 43
“I used a uniform density value for kappa (Gelman, 2006) that ranged from 0.”
MethodPopulation‐level inference for home‐range areas
Fleming, C. H., Deznabi, I., Alavi, S. E., et al. (2022) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.13815 · Cited by 44
“Finally, our Bayesian analysis produced much larger small- m biases that we anticipated, even though we supplied a non- informative prior similar to that suggested by Gelman (2006) for variance parameters, and further assisted our Bayesian analysis by fixing each prior’s mode to the truth.”
ResultRoy, S., & Attri, R. (2022) · Tourism Management · doi:10.1016/j.tourman.2022.104544 · Cited by 53
Hierarchical analysis of multiple noisy abundance indices
Conn, P. B. (2009) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/f09-175 · Cited by 46
“Finally, for spi , Gelman (2006) suggests that a Uniform(0, m) distribution may outperform other choices when there is a small number of group effects.”
BackgroundEvidential Strength of Intonational Cues and Rational Adaptation to (Un‐)Reliable Intonation
Roettger, T. B., & Franke, M. (2019) · Cognitive Science · doi:10.1111/cogs.12745 · Cited by 51
Trends and Carrying Capacity of Sea Otters in Southeast Alaska
Tinker, M. T., Gill, V. A., Esslinger, G. G., Bodkin, J. L., Monk, M., Mangel, M., Monson, D. H., Raymond, W. W., & Kissling, M. L. (2019) · Journal of Wildlife Management · doi:10.1002/jwmg.21685 · Cited by 55
Mod, H. K., Chevalier, M., Luoto, M., & Guisan, A. (2020) · Journal of Ecology · doi:10.1111/1365-2745.13434 · Cited by 60
“We used half-Cauchy distributions for standard deviation parameters associated with random effects on sites (Gelman, 2006).”
MethodImportance Sampling Squared for Bayesian Inference in Latent Variable ModelsPreprint
Tran, M., Scharth, M., Pitt, M. K., & Kohn, R. (2014) · SSRN Electronic Journal · doi:10.2139/ssrn.2386371 · Cited by 55
Buczinski, S., Fecteau, G., Dubuc, J., & Francoz, D. (2018) · Preventive Veterinary Medicine · doi:10.1016/j.prevetmed.2018.05.004 · Cited by 43
“The specific herd term (ε) accounted for data structure with calves clustered within specific farms: (4) ε∼ norm (0;τ) and τ∼ dgamma (1,1) The choice of gamma (γ=1, γ=1) for precision specification was considered as reasonably non-informative (Gelman, 2006) with γ≤ 1 avoiding γ→ 0 (eg 0.”
BackgroundInfluentialCeccherini, G., Girardello, M., Beck, P. S. A., Migliavacca, M., Duveiller, G., Dubois, G., Avitabile, V., Battistella, L., Barredo, J. I., & Cescatti, A. (2023) · Communications Earth & Environment · doi:10.1038/s43247-023-00758-w · Cited by 52
Bayesian hierarchical modeling: an introduction and reassessment
Veenman, M., Stefan, A. M., & Haaf, J. M. (2023) · Behavior Research Methods · doi:10.3758/s13428-023-02204-3 · Cited by 77
“Wide priors are priors that allocate (approximately) equal probability density to a great range of values, such as a uniform distribution without bounds (Gelman, 2006).”
BackgroundGrey seal predation impairs recovery of an over‐exploited fish stock
Cook, R., Holmes, S. J., & Fryer, R. J. (2015) · Journal of Applied Ecology · doi:10.1111/1365-2664.12439 · Cited by 64
Default Bayesian model determination methods for generalised linear mixed models
Overstall, A. M., & Forster, J. J. (2010) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2010.03.008 · Cited by 40
“Gelman (2006) points out that inverse-gamma (and inverse-Wishart) prior distributions for variance components can be overly informative, even with “non-informative” choices for the hyperparameters.”
BackgroundThe heritability of chimpanzee and human brain asymmetry
Gómez‐Robles, A., Hopkins, W. D., Schapiro, S. J., & Sherwood, C. C. (2016) · Proceedings of the Royal Society B Biological Sciences · doi:10.1098/rspb.2016.1319 · Cited by 46
Variance prior specification for a basket trial design using Bayesian hierarchical modeling
Cunanan, K. M., Iasonos, A. E., Shen, R., & Gönen, M. (2018) · Clinical Trials · doi:10.1177/1740774518812779 · Cited by 46
“This is a challenge that is particular to basket trials; most other applications of hierarchical models will have several (in some cases hundreds of) random effects.(18) There is also the issue that the variance parameter in hierarchical models is central to the questions posed by a basket trial, whereas in many appli…”
InfluentialMultiscale Characterization of Heart Failure
Costabal, F. S., Choy, J. S., Sack, K. L., Guccione, J. M., Kassab, G. S., & Kuhl, E. (2018) · Acta Biomaterialia · doi:10.1016/j.actbio.2018.12.053 · Cited by 44
Suchting, R., Colpo, G. D., Rocha, N. P., & Ahn, H. B. (2019) · Biological Research For Nursing · doi:10.1177/1099800419869845 · Cited by 48
“Second, Bayesian inference may be better equipped for examining small samples with appropriately informative priors (Gelman, 2006; McNeish, 2016).”
BackgroundZhang, X., Xu, C., & Xiao, G. (2020) · The Science of The Total Environment · doi:10.1016/j.scitotenv.2020.136623 · Cited by 43
“The term b1i, quantified the deviation from the overall temporal variation b0 (Li et al., 2014). εit was chosen to denote the Gaussian noise following the normal distribution N (0, σε2) (Gelman, 2006).”
BackgroundFitting Multilevel Vector Autoregressive Models in Stan, JAGS, and Mplus
Li, Y., Wood, J., Ji, L., Chow, S., & Oravecz, Z. (2021) · Structural Equation Modeling A Multidisciplinary Journal · doi:10.1080/10705511.2021.1911657 · Cited by 44
“Using conjugate priors has some known advantages (e.g., familiarity and tractability of the forms of the posterior distributions), but also some known drawbacks in the context of multilevel modeling (Gelman et al., 2006).”
Fixing the stimulus-as-fixed-effect fallacy in task fMRI
Westfall, J., Nichols, T. E., & Yarkoni, T. (2016) · Wellcome Open Research · doi:10.12688/wellcomeopenres.10298.2 · Cited by 92
BAYSIAN INFERENCE FOR ORDERED RESPONSE DATA WITH A DYNAMIC SPATIAL‐ORDERED PROBIT MODEL
Wang, X., & Kockelman, K. M. (2009) · Journal of Regional Science · doi:10.1111/j.1467-9787.2009.00622.x · Cited by 43
“(38) In other words, ( ) ∝ 1,(39) 3In a recent study, Gelman (2006) suggests that this inverse gamma prior may lead to an improper posterior distribution.”
BackgroundBayesian Analysis in Stata with WinBUGS
Thompson, J., Palmer, T., & Moreno, S. G. (2006) · The Stata Journal Promoting communications on statistics and Stata · doi:10.1177/1536867x0600600406 · Cited by 43
Barceló, M. A., Sáez, M., Serral, G., Martínez‐Beneito, M. A., Martínez, J. M., Borrell, C., Ocaña‐Riola, R., Montoya, I., Calvo, M., López‐Abente, G., et al. (2008) · Gaceta Sanitaria · doi:10.1016/s0213-9111(08)75362-7 · Cited by 46
A Bayesian approach to multiscale inverse problems with on-the-fly scale determination
Ellam, L., Zabaras, N. J., & Girolami, M. (2016) · Journal of Computational Physics · doi:10.1016/j.jcp.2016.08.031 · Cited by 42
“(36), we follow [32] and adopt the uninformative choices of αe = βe = 0.”
MethodBaySTDetect: detecting unusual temporal patterns in small area data via Bayesian model choice
Li, G., Best, N., Hansell, A., Ahmed, I., & Richardson, S. T. (2012) · Biostatistics · doi:10.1093/biostatistics/kxs005 · Cited by 55
Risk Factors for Foot-and-Mouth Disease in Tanzania, 2001-2006
Allepuz, A., Stevenson, M. A., Kivaria, F. M., Berkvens, D. L., Casal, J., & Picado, A. (2013) · Transboundary and Emerging Diseases · doi:10.1111/tbed.12087 · Cited by 43
Comparison of Statistical and Deterministic Frameworks of Uncertainty Quantification
Frenklach, M., Packard, A., García‐Donato, G., Paulo, R., & Sacks, J. (2016) · SIAM/ASA Journal on Uncertainty Quantification · doi:10.1137/15m1019131 · Cited by 39
Kallasvuo, M., Vanhatalo, J. P., & Veneranta, L. (2016) · Canadian Journal of Fisheries and Aquatic Sciences · doi:10.1139/cjfas-2016-0008 · Cited by 42
“290 We used log-uniform priors for the variance parameters ;F*, ;H*, ;*̀ and 291 p;@A,_* , ;@A* q/120 and weakly informative half Student-t priors (Gelman 2006) for the other 292 hyper-parameters.”
MethodZhang, J., & Shields, M. D. (2018) · Computer Methods in Applied Mechanics and Engineering · doi:10.1016/j.cma.2018.01.045 · Cited by 40
“It is worth noting that a diffuse or vague prior may not be uniform and sometimes a diffuse prior can be more informative than the uniform prior [54, 55, 56].”
BackgroundDelia, J. R. J., Bravo‐Valencia, L., & Warkentin, K. M. (2020) · Ecological Monographs · doi:10.1002/ecm.1411 · Cited by 48
Detection of atmospheric rivers with inline uncertainty quantification: TECA-BARD v1.0.1
O'Brien, T. A., Risser, M. D., Loring, B., Elbashandy, A., Krishnan, H., Johnson, J., Patricola, C. M., O'Brien, J. P., Mahesh, A., Prabhat, et al. (2020) · Geoscientific model development · doi:10.5194/gmd-13-6131-2020 · Cited by 52
Weak Informativity and the Information in One Prior Relative to Another
Evans, M. J., & Jang, G. H. (2012) · Cited by 42
Douglas, J., Mendes, F. K., Bouckaert, R., Xie, D., Jiménez-Silva, C. L., Swanepoel, C., de Ligt, J., Ren, X., Storey, M., Hadfield, J., et al. (2021) · Virus Evolution · doi:10.1093/ve/veab052 · Cited by 39
Patxot, M., Baños, D. T., Kousathanas, A., Orliac, E. J., Ojavee, S. E., Moser, G., Holloway, A., Sidorenko, J., Kutalik, Z., Mägi, R., et al. (2021) · Nature Communications · doi:10.1038/s41467-021-27258-9 · Cited by 45
The inclusion of historical control data may reduce the power of a confirmatory study
Cuffe, R. L. (2011) · Statistics in Medicine · doi:10.1002/sim.4212 · Cited by 46
Nonparametric Priors for Ordinal Bayesian Social Science Models: Specification and Estimation
Gill, J., & Casella, G. (2009) · Journal of the American Statistical Association · doi:10.1198/jasa.2009.0039 · Cited by 42
Almaraz, P., & Oró, D. (2011) · Ecology · doi:10.1890/11-0181.1 · Cited by 45
A Bayesian Approach to More Stable Estimates of Group-Level Effects in Contextual Studies
Zitzmann, S., Lüdtke, O., & Robitzsch, A. (2015) · Multivariate Behavioral Research · doi:10.1080/00273171.2015.1090899 · Cited by 48
“Another prior distribution suggested for SDs is a uniform distribution that ranges from 0 to A (see Gelman, 2006; Natarajan & Kass, 2000): τX ∼ U (0, A) , A > 0, (14) where a smaller scale parameter A implies a more restrictive distribution.”
BackgroundLurier, T., Rousset, É., Gasqui, P., Sala, C., Claustre, C., Abrial, D., Dufour, P., CRÉMOUX, R. D., Gache, K., Delignette‐Muller, M. L., et al. (2021) · Veterinary Research · doi:10.1186/s13567-021-00926-w · Cited by 40
Multicohort analysis of the maternal age effect on recombination
Martin, H. C., Christ, R., Hussin, J., O’Connell, J., Gordon, S. D., Mbarek, H., Hottenga, J., McAloney, K., Willemsen, G. A. H. M., Gasparini, P., et al. (2015) · Nature Communications · doi:10.1038/ncomms8846 · Cited by 40
Prior Effective Sample Size in Conditionally Independent Hierarchical Models
Morita, S., Thall, P. F., & Müller, P. (2012) · Bayesian Analysis · doi:10.1214/12-ba720 · Cited by 44
Markov chain Monte Carlo estimation of a mixture item response theory model
Cho, S., Cohen, A. S., & Kim, S. (2011) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2011.603090 · Cited by 41
Age Differences in Long-Term Patterns of Change in Alcohol Consumption Among Aging Adults
Shaw, B. A., KRAUSE, N. M., Liang, J., & McGeever, K. (2010) · Journal of Aging and Health · doi:10.1177/0898264310381276 · Cited by 46
“First, although there is a consensus for weighting data in generating descriptive statistics for a given target population, there is no such agreement in multivariate analyses (Gelman, 2006).”
BackgroundBounthavong, M., Zargarzadeh, A. H., Hsu, D. I., & Vanness, D. J. (2011) · Value in Health · doi:10.1016/j.jval.2010.12.006 · Cited by 43
A Bayesian Framework for Inference of the Genotype–Phenotype Map for Segregating Populations
Hageman, R. S., Leduc, M. S., Korstanje, R., Paigen, B., & Churchill, G. A. (2011) · Genetics · doi:10.1534/genetics.110.123273 · Cited by 41
“Notably, as a, b / 0, the posterior does not have a proper limiting distribution, and posterior inference has been shown to be sensitive (Dongen 2006; Gelman 2006).”
BackgroundWerner, J., Luterbacher, J., & Smerdon, J. E. (2013) · Journal of Climate · doi:10.1175/jcli-d-12-00016.1 · Cited by 42
“The selected hyperparameters are shown in Table A1 in the appendix in the discussion of the convergence of the algorithm: cf. Gelman (2006) for a discussion on the subtleties of prior selection in hierarchical models.”
BackgroundHickey, G. L., Craig, P. S., Luttik, R., et al. (2012) · Environmental Toxicology and Chemistry · doi:10.1002/etc.1891 · Cited by 41
Hierarchical Bayesian fatigue data analysis
Liu, X., Lü, D., & Hoogenboom, P. C. J. (2017) · International Journal of Fatigue · doi:10.1016/j.ijfatigue.2017.03.043 · Cited by 41
“Gelman also recommended starting with a noninformative uniform prior density Uð0;uÞ on a wide range of standard deviation parameters in fitting hierarchical models.”
BackgroundInfluentialPedroza, C., & Truong, V. T. T. (2017) · Trials · doi:10.1186/s13063-017-2248-1 · Cited by 47
“fitted the Bayesian GLMM using vague Normal(0,10(4)) priors for β0, β1, and β2 and half-Cauchy(0,1) [33] for σ.”
MethodA unified view on Bayesian varying coefficient models
Franco‐Villoria, M., Ventrucci, M., & Rue, H. (2019) · Electronic Journal of Statistics · doi:10.1214/19-ejs1653 · Cited by 44
Fayard, M., Dechaume‐Moncharmont, F., Wattier, R. A., & Perrot‐Minnot, M. (2020) · Biological reviews/Biological reviews of the Cambridge Philosophical Society · doi:10.1111/brv.12606 · Cited by 47
“The analyses were based on Bayesian hierarchical models which impose definitions of priors (Gelman, 2006).”
MethodStability in Activity and Boldness Across Time and Context in Captive Siberian Dwarf Hamsters
Kanda, L. L., Louon, L., & Straley, K. (2012) · Ethology · doi:10.1111/j.1439-0310.2012.02038.x · Cited by 42
Expected time-invariant effects of biological traits on mammal species duration
Smits, P. D. (2015) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1510482112 · Cited by 43
Nandram, B., & Choi, J. W. (2010) · Journal of the American Statistical Association · doi:10.1198/jasa.2009.ap08443 · Cited by 45
“(20) Finally, a priori, we assume that β and σ 2 are independent with β ∼ Normal(θ0, 0) and (21) p(σ 2) = 1/(1 + σ 2)2, σ 2 > 0, where θ0 and 0 are specified as done by Nandram and Choi (2005), thereby avoiding the ill-behaved prior distribution σ−2 ∼ Gamma(a,a) with a = 0.001 (see Gelman 2006).”
BackgroundInfluentialGeneralized fiducial inference for normal linear mixed models
Cisewski-Kehe, J., & Hannig, J. (2012) · The Annals of Statistics · doi:10.1214/12-aos1030 · Cited by 53
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.