Prior distributions for variance parameters in hierarchical models (comment on article by Browne and Draper)
Open accessThe DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index; 1 of the 23 sources its claims cite were checked for support.
Abstract
Various noninformative prior distributions have been suggested for scale parameters in hierarchical models. We construct a new folded-noncentral-$t$ family of conditionally conjugate priors for hierarchical standard deviation parameters, and then consider noninformative and weakly informative priors in this family. We use an example to illustrate serious problems with the inverse-gamma family of "noninformative" prior distributions. We suggest instead to use a uniform prior on the hierarchical standard deviation, using the half-$t$ family when the number of groups is small and in other settings where a weakly informative prior is desired. We also illustrate the use of the half-$t$ family for hierarchical modeling of multiple variance parameters such as arise in the analysis of variance.
Overview
CoChat Summary
TL;DR
Gelman argues that inverse-gamma “noninformative” priors for hierarchical variance parameters can badly distort posterior inference, especially when the group-level variance is near zero or the number of groups is small. He recommends a uniform prior on the hierarchical standard deviation when groups are not few, and half-t/half-Cauchy weakly informative priors otherwise, developed within a folded-noncentral-t conditionally conjugate family.
Key findings
- The inverse-gamma(ε, ε) family is not reliably noninformative: as ε → 0 it yields an improper posterior, and in the 8-schools example inverse-gamma(1, 1) and inverse-gamma(0.001, 0.001) priors on σ²_α constrained posterior inferences and increased shrinkage relative to a uniform prior on σ_α.
- A uniform prior on σ_α performed well for the 8-schools data (J = 8), but with only J = 3 schools it produced an extremely long right tail and undershrinkage; a half-Cauchy prior with scale A = 25 removed much of that unrealistic upper tail.
- For multiple variance components, Gelman recommends a hierarchical half-Cauchy model with scale estimated from data; in the 5 × 5 × 2 split-plot latin square Anova with nine variance components, the Cauchy scale parameter A was estimated at 1.8 with a 95% posterior interval of [0.5, 5.1].
- The folded-noncentral-t family, obtained via a redundant multiplicative parameterization, provides conditionally conjugate priors for σ_α and includes the improper uniform density (ν = −1) and proper half-Cauchy (ν = 1) as special cases.
Methods
- Conceptual/comment paper using a simple two-level normal hierarchical model: y_ij ∼ N(μ + α_j, σ²_y), α_j ∼ N(0, σ²_α), focusing on the prior for σ_α.
- Illustrates with the 8-schools example (J = 8, 6000 Bugs simulation draws), a 3-schools subset, and a 5 × 5 × 2 split-plot latin square Anova with nine variance components.
- Uses a parameter-expanded form y_ij ∼ N(μ + ξη_j, σ²_y), η_j ∼ N(0, σ²_η), with σ_α = |ξ|σ_η, to construct and compare uniform, inverse-gamma, half-Cauchy, and hierarchical half-Cauchy priors.
Limitations
- The paper states that noninformative and weakly informative priors are inherently provisional: after fitting, one should inspect the posterior and revise the prior if it does not make sense.
- The paper notes that the uniform prior on σ_α has a positive miscalibration and yields an improper posterior for J = 1 or 2; inverse-gamma inferences are sensitive to ε; and half-Cauchy scale choices are weakly informative rather than fully objective.
- Evident from the text: the recommendations rest on illustrative examples and conceptual argument rather than a broad simulation study or systematic empirical comparison, and extensions to nonlinear, multivariate, and more complex hierarchical models are left for further work.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 Valid · 1 Not valid · 14 Not checked · 56 Own results and interpretations
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Similarly, there are no always-nonnegative classical unbiased estimators of σ α or σ 2 α in the hierarchical model.”
- Also stated in: § 7.1 Prior distributions for variance parameters · plain
“Similarly, there are no alwaysnonnegative classical unbiased estimators of σ α or σ 2 α in the hierarchical model.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 2.4 Calibration
What the sources say
- Could not verifycites Bickel and Blackwell (1967) matched to “A Note on Bayes Estimates” (1967) sourceExperimental — accuracy not yet measured
The introduction (p. 1) and Theorem 1.3 (p. 3) concern when an unbiased Bayes estimate must have zero Bayes risk. Searches for hierarchical models, σₐ, σ²ₐ, and always-nonnegative classical unbiased estimators found no passage addressing the claimed nonexistence result. (LLM, conf=0.77)
- Not checked — no full text of the sourcecites Meng and Zaslavsky (2002) matched to “Single observation unbiased priors” (2002) source
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 7.1 Prior distributions for variance parameters · plain
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“If J is low, the uniform prior density tends to lead to high estimates of σ α , as discussed in Section 5.2.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 7.1 Prior distributions for variance parameters
hedgedstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In contrast, the hierarchical half-Cauchy model performs a great deal of shrinkage, especially of the high ranges of the intervals.”
- Also stated in: § Bugs model: a half-Cauchy prior distribution on sigma.theta is induced # using a normal prior on xi and an inverse-gamma on tau.eta model { for (j in 1:J){ # J = the number of schools y[j] ~dnorm (theta[j], tau.y[j]) # data model: the likelihood theta[j] <-mu.theta+ xi*eta[j] tau.y[j] <-pow(sigma.y[j],-2) } xi ~dnorm (0, tau.xi) tau.xi <-pow(prior.scale,-2) for (j in 1:J){ eta[j] ~dnorm (0, tau.eta) # hierarchical model for theta } tau.eta ~dgamma (.5, .5)# chi^2 with 1 d.f.sigma.theta<-abs(xi)/sqrt(tau.eta)# cauchy = normal/sqrt(chi^2) mu.theta ~dnorm (0.0, 1.0E-6) # noninformative prior on mu } · plain
“The half-Cauchy model gives much sharper inferences, using the partial pooling that comes with fitting a hierarchical model.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 6.2 Superpopulation and finite-population standard deviations
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § Bugs model: a half-Cauchy prior distribution on sigma.theta is induced # using a normal prior on xi and an inverse-gamma on tau.eta model { for (j in 1:J){ # J = the number of schools y[j] ~dnorm (theta[j], tau.y[j]) # data model: the likelihood theta[j] <-mu.theta+ xi*eta[j] tau.y[j] <-pow(sigma.y[j],-2) } xi ~dnorm (0, tau.xi) tau.xi <-pow(prior.scale,-2) for (j in 1:J){ eta[j] ~dnorm (0, tau.eta) # hierarchical model for theta } tau.eta ~dgamma (.5, .5)# chi^2 with 1 d.f.sigma.theta<-abs(xi)/sqrt(tau.eta)# cauchy = normal/sqrt(chi^2) mu.theta ~dnorm (0.0, 1.0E-6) # noninformative prior on mu } · plain
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This prior distribution is even more sharply peaked near zero and further distorts posterior inferences, with the problem arising because the marginal likelihood for σ α remains high near zero.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 5.1 Noninformative prior distributions for the 8-schools problem
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The uniform prior distribution seems fine for the 8-school analysis”
- Also stated in: § 5.2 Weakly informative prior distribution for the 3-schools problem · plain
“the default uniform prior gave reasonable results”
- Also stated in: § 7.1 Prior distributions for variance parameters · hedged
“Figure 1 illustrates the generally robust properties of the uniform prior density on σ α .”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 5.2 Weakly informative prior distribution for the 3-schools problem
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Also stated in: § 5.2 Weakly informative prior distribution for the 3-schools problem · plain
26 more claims
Create a free account to see all 71 claims and every source verdict.
- It may state two separate claims as one. 34
- Probably not a substantial claim. 12
- The quote only partly states this claim. 1
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Because the uniform distribution tends to unduly favor presence of heterogeneity in discrimination and calibration estimates across studies,58,59”
From A framework for meta-analysis of prediction model studies with binary and time-to-event outcomes (2019) · cites this paper as [59]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 7.1 Prior distributions for variance parameters
“we see conditional conjugacy in the wider class of half-t distributions on σ α , which include the uniform and half-Cauchy densities on σ α (as well as inverse-gamma on σ 2 α ) as special cases.”
§ Prior distributions for variance parameters in hierarchical models(Comment on Article by Browne and Draper) > 3.2 Folded-noncentral-t prior distribution for σ α
“In the limit that the denominator is specified exactly, we have a folded normal distribution; conversely, specifying the numerator exactly yields the square-root-inverse-χ 2 distribution for σ α , as in Section 3.1.”
Citations
Cites
Gelman, A., Carlin, J. B., Stern, H. S., & Rubin, D. B. (2003) · doi:10.1201/9780429258480 · Cited by 27,276
Self-citationKass, R. E., & Raftery, A. E. (1995) · Journal of the American Statistical Association · doi:10.1080/01621459.1995.10476572 · Cited by 12,558
Sampling-based approaches to calculating marginal densities
Gelfand, A. E., & Smith, A. F. M. (1990) · Journal of the American Statistical Association · doi:10.1080/01621459.1990.10476213 · Cited by 6,606 · cited 1×
“Typically, enough data will be available to estimate µ and σy that one can use any reasonable noninformative prior distribution—for example, p(µ, σy) ∝ 1 or p(µ, log σy) ∝ 1.”
Introducing Multilevel Modeling
Kreft, I., & de Leeuw, J. (1998) · doi:10.4135/9781849209366 · Cited by 2,997
POSTERIOR PREDICTIVE ASSESSMENT OF MODEL FITNESS VIA REALIZED DISCREPANCIES
Gelman, A., Meng, X., & Stern, H. S. (1996) · Cited by 2,638
Self-citationBayes and Empirical Bayes Methods for Data Analysis
(2001) · Technometrics · doi:10.1198/tech.2001.s608 · Cited by 2,098
INADMISSIBILITY OF THE USUAL ESTIMATOR FOR THE MEAN OF A MULTIVARIATE NORMAL DISTRIBUTION
Stein, C. (1956) · Contribution to the Theory of Statistics · doi:10.1525/9780520313880-018 · Cited by 1,841
Bayesian Approaches to Clinical Trials and Health‐Care Evaluation
Spiegelhalter, D. J., Abrams, K. R., & Myles, J. P. (2003) · Wiley eBooks · doi:10.1002/0470092602 · Cited by 1,673
Parametric empirical Bayes inference: theory and applications
Morris, C. N. (1983) · Journal of the American Statistical Association · doi:10.1080/01621459.1983.10477920 · Cited by 1,448 · cited 1×
“On the theoretical side, hierarchical models allow a more “objective” approach to inference by estimating the parameters of prior distributions from data rather than requiring them to be specified using subjective information (see James and Stein, 1960, Efron and Morris, 1975, and Morris, 1983).”
The Selection of Prior Distributions by Formal Rules
Kass, R. E., & Wasserman, L. (1996) · Journal of the American Statistical Association · doi:10.1080/01621459.1996.10477003 · Cited by 1,308
Research works printed but not linked: 5
- Box, G. E. P. and Tiao, G. C. (1973). Bayesian Inference in Statistical Analysis. Reading, Mass.: Addison-Wesley.
- James, W. and Stein, C. (1960). "Estimation with quadratic loss." In Neyman, J. (ed.), Proceedings of the Fourth Berkeley Symposium, volume 1, 361–380. Berkeley: University of California Press.
- Johnson, N. L. and Kotz, S. (1972). Distributions in Statistics. New York: Wiley. 4 vols.
- Liu, C. (2001). "Bayesian analysis of multivariate probit models. Discussion of “The art of data augmentation” by D. A. van Dyk and X. L. Meng." Journal of Computational and Graphical Statistics, 10: 75–81.
- Raudenbush, S. W. and Bryk, A. S. (2002). Hierarchical Linear Models. Thousand Oaks, Calif.: Sage., second edition.
Cited by
Items 2,201–2,300 of 4,089
Bayesian semiparametric copula estimation with application to psychiatric genetics
Rosen, O., & Thompson, W. K. (2015) · Biometrical Journal · doi:10.1002/bimj.201300130 · Cited by 6
“Priors on the τ js The τ js are assumed independent Half-t(ν j, Aj ) (Gelman, 2006) with pdf p(x) ∝ [1 + (x/Aj )2/ν j )]−(ν j+1)/2, x > 0, where the hyperparameters ν j and Aj are assumed known.”
BackgroundFlexibility of Bayesian generalized linear mixed models for oral health research
Berchialla, P., Baldi, I., Notaro, V., Barone‐Monfrin, S., Bassi, F., & Gregori, D. (2009) · Statistics in Medicine · doi:10.1002/sim.3648 · Cited by 11
Non‐inferiority and networks: inferring efficacy from a web of data
Lin, J., Gamalo, M., & Tiwari, R. C. (2015) · Pharmaceutical Statistics · doi:10.1002/pst.1729 · Cited by 8
“For instance, [28] recommends against using a uniform prior because it tends to unduly favor higher values, resulting in excessive heterogeneity.”
BackgroundBayesian change points analysis for earthquakes body wave magnitude
Al-Awadhi, F. A., & Al-Hulail, D. (2016) · Journal of Applied Statistics · doi:10.1080/02664763.2015.1117585 · Cited by 7
Themelis, K. E., Rontogiannis, A. A., & Koutroumbas, K. (2016) · IEEE Transactions on Signal Processing · doi:10.1109/tsp.2016.2543204 · Cited by 7
Choi, J., & O’Malley, A. J. (2016) · Journal of the Royal Statistical Society Series C (Applied Statistics) · doi:10.1111/rssc.12158 · Cited by 7
“It has been noted that IG priors with large variances can be problematic when used for variance parameters at levels of the model above the observation level (Gelman, 2006).”
BackgroundSpatiotemporal Trends in Teen Birth Rates in the USA, 2003–2012
Khan, D., Rossen, L. M., Hamilton, B. E., Dienes, E., He, Y., & Wei, R. X. (2017) · Journal of the Royal Statistical Society Series A (Statistics in Society) · doi:10.1111/rssa.12266 · Cited by 7
“0:01, 0:01/ (Gelman, 2006) resulted in the lowest deviance information criterion (DIC) in model comparisons (the DIC is described in detail in Section 3.”
InfluentialMorgan, J. W., Vincent, J. D., & Camac, J. (2017) · Journal of Plant Ecology · doi:10.1093/jpe/rtw130 · Cited by 7
“Standard deviations for observation, σobs, transect, σtran, and individual σind were estimated from positive half Cauchy priors with a scale of 25 (Gelman 2006).”
MethodMedical overpayment estimation: A Bayesian approach
Musal, R. M., & Ekin, T. (2017) · Statistical Modelling · doi:10.1177/1471082x16685020 · Cited by 7
Lu, T. (2017) · Statistical Methods in Medical Research · doi:10.1177/0962280217704225 · Cited by 7
Flexible Mixture-Amount Models Using Multivariate Gaussian Processes
Ruseckaite, A., Fok, D., & Goos, P. (2018) · Journal of Business and Economic Statistics · doi:10.1080/07350015.2018.1497506 · Cited by 7
Navas, A. L. A., Osei, F., Leonardo, L., et al. (2019) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph16020176 · Cited by 7
“Half‐normal was selected in order to restrict our prior to positive values and avoid problems with convergence [52, 56].”
BackgroundClaveria, R. M., Mendez, R. A., Silva, J. F., & Orchard, M. E. (2019) · Publications of the Astronomical Society of the Pacific · doi:10.1088/1538-3873/ab22e2 · Cited by 8
Semiparametric covariate-modulated local false discovery rate for genome-wide association studies
Zablocki, R., Levine, R. A., Schork, A. J., Xu, S., Wang, Y., Fan, C. C., & Thompson, W. K. (2017) · The Annals of Applied Statistics · doi:10.1214/17-aoas1077 · Cited by 9
Kurz, C. F., & Hatfield, L. A. (2018) · bioRxiv · doi:10.1002/sim.8307 · Cited by 7
“[52] As before, this mixture model produces posterior estimates of the prevalence of each component and the parameters that govern it, plus posterior probabilities of membership in each mixture component for each observation.”
Liu, H., Zhu, X., Zhang, D., et al. (2019) · ISPRS International Journal of Geo-Information · doi:10.3390/ijgi8110488 · Cited by 7
Conventional and unconventional participation in Latin America: a hierarchical latent class approach
Alvarez, R. M., Katz, G., Levin, I., & Núñez, L. (2020) · Political Science Research and Methods · doi:10.1017/psrm.2020.35 · Cited by 8
“In contrast, previous research (e.g., Gelman, 2006) has demonstrated that Bayesian methods yield accurate estimates of the regression parameters and variance components of hierarchical models even with a small number of clusters, provided the number of observations per cluster is reasonably…”
ResultIntegrative statistical methods for exposure mixtures and health
Reich, B. J., Guan, Y., Fourches, D., Warren, J. L., Sarnat, S., & Chang, H. H. (2020) · The Annals of Applied Statistics · doi:10.1214/20-aoas1364 · Cited by 8
“The remaining parameters have uninformative priors α ∼ Normal(0, c21I), b, γ01, γ02 ∼ Normal(0, c21), and σ, τ1, τ2 ∼ HC(c2), i.e., the half-Cauchy prior with scale parameter c2 (Gelman, 2006).”
Geda, M. W., & Kwong, C. K. (2021) · Journal of remanufacturing · doi:10.1007/s13243-020-00099-3 · Cited by 10
The Electoral System and Women's Legislative Underrepresentation in Post-Communist Ukraine
Thames, F. C. (2018) · Comparative Politics · doi:10.5129/001041518822263656 · Cited by 7
Rhodes, S., Buchsbaum, B. R., & Hasher, L. (2021) · Psychonomic Bulletin & Review · doi:10.3758/s13423-021-01981-2 · Cited by 21
“For random (individual-level) effects we used a half-Cauchy (0, 2.5) prior on standard deviations (Gelman, 2006) and an LKJ (1) prior (Lewandowski et al., 2009) on correlation matrices (i.e., a uniform prior).”
MethodInfluentialBayesian Analysis of Anova and Mixed Models on the Log-Transformed Response Variable
Gardini, A., Trivisano, C., & Fabrizi, E. (2021) · Psychometrika · doi:10.1007/s11336-021-09769-y · Cited by 9
“This special case is relevant to discuss the uniform prior on the standard deviation advocated by Gelman (2006).”
Yu, Y., Todd, J. T., & Petrov, A. A. (2021) · Journal of Vision · doi:10.1167/jov.21.6.5 · Cited by 9
“We placed weakly informative Gaussian priors on the group-means other than μθk , and placed noninformative priors with large range, Uniform (0, 10), on the group-level standard deviations, as suggested by Gelman (2006).”
MethodRoos, M., Hunanyan, S., Bakka, H. C., & Rue, H. (2021) · Biometrical Journal · doi:10.1002/bimj.202000193 · Cited by 7
“These results show that the HC heterogeneity prior (Gelman, 2006; Polson & Scott, 2012; Gelman et al., 2017) is a valid alternative to HN, which is often recommended in the literature (Bender et al., 2018; Friede et al., 2017a, 2017b).”
BackgroundMethodInfluentialHunger state and not personality determines task participation in a spider society
Parthasarathy, B., Müller, M., Bilde, T., & Schneider, J. M. (2022) · Animal Behaviour · doi:10.1016/j.anbehav.2022.06.002 · Cited by 8
Berrett, C., Williams, G. P., Moon, T. K., & Gunther, J. H. (2014) · Technometrics · doi:10.1080/00401706.2013.869262 · Cited by 8
“Finally, following Gelman (2006), we use an inverse gamma distribution for σ 2.”
MethodFlexible von Bertalanffy growth models incorporating Bayesian splines
Chambers, M. S., Sidhu, L. A., O’Neill, B., et al. (2017) · Ecological Modelling · doi:10.1016/j.ecolmodel.2017.03.026 · Cited by 7
Tessema, T. (2017) · Annals of Data Science · doi:10.1007/s40745-017-0121-4 · Cited by 7
Spatial Panel Data Model with Error Dependence: A Bayesian Separable Covariance Approach
Leorato, S., & Mezzetti, M. (2015) · Bayesian Analysis · doi:10.1214/15-ba979 · Cited by 7
The Bayesian Multilevel Trifactor Item Response Theory Model
Fujimoto, K. (2018) · Educational and Psychological Measurement · doi:10.1177/0013164418806694 · Cited by 6
“…that is, ~su1;half -t(0, 1, 1): ð5Þ The choice of a half-t distribution was because research has suggested that this distribution leads to a better behaving Markov chain for the SD when the SD is near 0, compared to when an inverse-gamma distribution is assigned on the variance (Gelman, 2006).”
BackgroundWuthrich, C., MacLatchy, L., & Nengo, I. O. (2019) · Scientific Reports · doi:10.1038/s41598-019-39800-3 · Cited by 7
Bayesian Analysis of a Physiologically Based Pharmacokinetic Model for Perchloroethylene in Humans
Qiu, J., Chien, Y., Bruckner, J. V., & Fisher, J. W. (2009) · Journal of Toxicology and Environmental Health · doi:10.1080/15287390903249099 · Cited by 7
“Half-Cauchy prior distributions, a special case of the conditionally conjugate folded-noncentral-t distributions, are recommended by Gelman (2006), but are not defined in MCSim 5.0.”
MethodResultInfluentialFalissard, B., Izard, V., Xerri, B., Bouvenot, G., Meyer, F., & Degos, L. (2010) · International Journal of Technology Assessment in Health Care · doi:10.1017/s0266462309990821 · Cited by 7
“) (9) and the arbitrary a priori distributions in Bayesian models which can impact on results (8).”
BackgroundTroude, P., Ancelet, S., Guibert, J., et al. (2012) · BMC Pregnancy and Childbirth · doi:10.1186/1471-2393-12-77 · Cited by 7
OKAMOTO, Y. (2012) · Japanese Psychological Research · doi:10.1111/j.1468-5884.2012.00513.x · Cited by 7
Creek habitats as sources for the spread of an invasive herb in a New Zealand mountain landscape.
Miller, A. L., Wiser, S. K., Sullivan, J. J., et al. (2015) · Lincoln University Research Archive (Lincoln University) · Cited by 7
Horseshoe Regularization for Feature Subset SelectionPreprint
Bhadra, A., Datta, J., Polson, N. G., et al. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1702.07400 · Cited by 7
“tacharya et al.,2015), to name a few. An open question is how these priors perform in terms of subset selection and whether fast computational algorithms are available. Following the recommendation ofGelman (2006), we used a standard half-Cauchy (C+(0,1)) prior in (2), similar to the original horseshoe formulation (Ca…”
MethodParameter estimation with a class of outer probability measuresPreprint
Houssineau, J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1801.00569 · Cited by 17
“omputation of the likelihood for highdimensional and complex data (Diggle and Gratton, 1984), the integration of non-random data (Diaconis and Zabell, 1982) and the definition of uninformative priors (Gelman, 2006). In this article, we show that including the modelling of deterministic uncertainty into standard probabi…”
Uncertainty Quantification in Multivariate Mixed Models for Mass Cytometry DataPreprint
Seiler, C., Kronstad, L. M., Simpson, L. J., Gars, M. L., Vendrame, E., Blish, C. A., & Holmes, S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.07976 · Cited by 13
Starling, J. E., Murray, J. S., Lohr, P. A., Aiken, A. R. A., Carvalho, C. M., & Scott, J. G. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.09405 · Cited by 7
“95, and default smoothing parameter κμ = 1, with a half-Cauchy prior on the scale of leaf parameters (Gelman, 2006).”
BackgroundMethodInfluentialMonitoring protocol development and assessment for narrowly endemic toads in Nevada, 2018
Halstead, B. J., Kleeman, P. M., Duarte, A., Rose, J. P., Urquhart, K. D., Mellison, C., Guadalupe, K. D., Cota, M., Horne, R. V., Killion, A., et al. (2019) · Antarctica A Keystone in a Changing World · doi:10.3133/ofr20191067 · Cited by 7
bamlss : A Lego Toolbox for Flexible Bayesian Regression (and Beyond)Preprint
Umlauf, N., Klein, N., Simon, T., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1909.11784 · Cited by 7
Slana, D. (2020) · eDiss (Georg-August-Universität Göttingen) · doi:10.53846/goediss-7942 · Cited by 6
Selecting relevant moderators with Bayesian regularized meta-regressionPreprint
Lissa, C. J. V., van Erp, S., & Clapper, E. (2021) · doi:10.31234/osf.io/6phs5 · Cited by 7
Development of a Bayesian multimodal model to detect biomarkers in neuroimaging studies
Bhaumik, D. K., Wang, Y., Yen, P., & Ajilore, O. (2023) · Frontiers in Neuroimaging · doi:10.3389/fnimg.2023.1147508 · Cited by 7
“Improper priors also yield improper posterior densities (Gelman, 2006; Gelman et al., 2013).”
BackgroundNakashima, D., Fujii, K., Kubo, Y., & Yorozuya, K. (2023) · Occupational Therapy International · doi:10.1155/2023/9505865 · Cited by 8
“The parameters were estimated using the Markov chain Monte Carlo method, and the halfCauchy distribution recommended as a weakly informative prior distribution was used for the prior distribution [29].”
BackgroundMethodROMI: a randomized two-stage basket trial design to optimize doses for multiple indications
Wang, S., Thall, P. F., Takeda, K., & Yuan, Y. (2024) · Biometrics · doi:10.1093/biomtc/ujae105 · Cited by 8
“Gelman ( 2006 ) and Chu and Yuan ( 2018b ) noted that the IG( (cid:8), (cid:8) ) with (cid:8) → 0 does not r epr e-se n t a non-inform ativ e prior, but in stead impos es strong shrink-age when the number of elements in the hierar chy (indica tions in our context) is small (eg, ≤ 6 ) unless the…”
Bayesian method for force identification within transfer path analysis
Shen, W., Schuhmacher, A., Song, W., Mediante, E. C., Lucklum, F., & Brunskog, J. (2025) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2025.112874 · Cited by 7
Wang, C., Sun, J., Liang, Z., Cai, Y., Zhang, M., Tian, S., Yang, W., Cheng, S., Duan, R., & Wang, X. (2025) · European Journal of Pain · doi:10.1002/ejp.70109 · Cited by 10
Technology mediation in child sexual exploitation and abuse in Africa and Asia
Ghai, S., Vuorre, M., Kardefelt‐Winther, D., Ferguson, A. M., Kurten, S., Livingstone, S., Przybylski, A. K., & Orben, A. C. (2026) · Nature · doi:10.1038/s41586-026-10525-4 · Cited by 8
Robust Estimation of Arps Decline Parameters Using Regularizing Priors
Ravikumar, A., & Lee, J. (2019) · SPE Eastern Regional Meeting · doi:10.2118/196576-ms · Cited by 6
A confidence interval robust to publication bias for random‐effects meta‐analysis of few studies
Henmi, M., Hattori, S., & Friede, T. (2021) · Research Synthesis Methods · doi:10.1002/jrsm.1482 · Cited by 7
“improper uniform, prior is used for the effect θ [25, 26].”
BackgroundMethodMizel, J. D., & Swanson, D. K. (2022) · Arctic Antarctic and Alpine Research · doi:10.1080/15230430.2022.2034373 · Cited by 7
“We specified vague normal priors for all regression coefficients and halfCauchy priors for the hierarchal standard deviation parameters (Gelman 2006).”
MethodBayesian Bootstrapped Correlation Coefficients
Rodriguez, J. E., & Williams, D. R. (2022) · The Quantitative Methods for Psychology · doi:10.20982/tqmp.18.1.p039 · Cited by 5
“…controlled by a single degrees of freedom parameter (Barnard et al., 2000), the marginal distribution for the variances have low density near zero (Gelman, 2006), and there is an a priori dependence between the resulting correlations and variances (Tokuda, Goodrich, Van Mechelen, Gelman, &…”
BackgroundMoral, R. D. A., Chen, Z., Zhang, S., McClean, S. I., Palma, G. R., Allan, B., & Kegel, I. (2022) · IEEE Access · doi:10.1109/access.2022.3215682 · Cited by 11
“…β k and φ c , since these parameters can take any real value; a flat gamma prior for ψ c , since these parameters are strictly positive; and half-Cauchy priors for the variance components, as suggested by [16], i.e. δ ∼ N(0 , 1000) Model implementation was carried out using JAGS [17] within R [18].”
MethodMuyters, G., & Maddens, B. (2023) · European Political Science Review · doi:10.1017/s1755773922000601 · Cited by 6
Russell, V. L., Stevens, M. H. H., Zeisler, A. A., & Ježková, T. (2021) · Journal of Mammalogy · doi:10.1093/jmammal/gyab126 · Cited by 6
Kemink, K. M., Diedrich, A., Adams, V. M., & Pressey, R. L. (2023) · Biological Conservation · doi:10.1016/j.biocon.2023.110193 · Cited by 7
A Bayesian Approach for Spatio-Temporal Data-Driven Dynamic Equation Discovery
North, J. S., Wikle, C. K., & Schliep, E. M. (2023) · Bayesian Analysis · doi:10.1214/23-ba1406 · Cited by 12
“Inference using traditional conjugate Inverse Gamma/Wishart priors are overly sensitive to the choice of hyperpriors when estimates are small (Gelman, 2006).”
MethodTerada, R., Matsuda, K., & Nishihara, G. N. (2023) · Phycological Research · doi:10.1111/pre.12539 · Cited by 6
“2015) and a Student ’ s prior distribution was placed on the scale parameter of the models (Gelman 2004, 2006).”
MethodCallum, K., Swinton, P. A., Gorely, T., Crabtree, D. R., & Leslie, S. J. (2023) · Heart & Lung · doi:10.1016/j.hrtlng.2023.12.002 · Cited by 9
Parvej, M., & Khan, A. A. (2024) · Journal of Applied Statistics · doi:10.1080/02664763.2024.2338404 · Cited by 9
“The use of a Half-Cauchy prior for various parameters is supported by [8], with detailed explanations provided by [6] and [11].”
Bargaining between the sexes: outside options and leisure time in hunter-gatherer households
Deb, A., Saunders, D., Major‐Smith, D., Dyble, M., Page, A. E., Salalι, G. D., Migliano, A. B., Heintz, C., & Chaudhary, N. (2024) · Evolution and Human Behavior · doi:10.1016/j.evolhumbehav.2024.05.003 · Cited by 7
Bayesian estimation of covariate assisted principal regression for brain functional connectivity
Park, H. (2024) · Biostatistics · doi:10.1093/biostatistics/kxae023 · Cited by 11
The Online Shortest Path Problem: Learning Travel Times Using a Multiarmed Bandit Framework
Lagos, T., Auad, R., & Lagos, F. (2024) · Transportation Science · doi:10.1287/trsc.2023.0196 · Cited by 6
Beddowes, E. J., Duran, M. O., Karapanagiotis, S., Martin, A. G., Gao, M., Masina, R., Woitek, R., Tanner, J., Tippin, F., Kane, J., et al. (2025) · Molecular Oncology · doi:10.1002/1878-0261.70015 · Cited by 8
“We used the BRMSR package version 2.14.4 to implement the models [37].”
InfluentialMultilevel Functional Principal Component Analysis of Façade Sound Insulation Data
Argiento, R., Bissiri, P. G., Pievatolo, A., & Scrosati, C. (2015) · Quality and Reliability Engineering International · doi:10.1002/qre.1843 · Cited by 6
Dual-Faceted Multidimensional IRT Models with Hierarchical Structure
Wang, L., & Finn, A. H. R. (2013) · Australasian Marketing Journal (AMJ) · doi:10.1016/j.ausmj.2013.02.002 · Cited by 6
Xing, L., Burstyn, I., Richardson, D. B., et al. (2013) · Statistics in Medicine · doi:10.1002/sim.5791 · Cited by 6
Bayesian covariance estimation and inference in latent Gaussian process models
Earls, C., & Hooker, G. (2013) · Statistical Methodology · doi:10.1016/j.stamet.2013.10.001 · Cited by 8
Song, H., Lawson, A. B., D’Agostino, R. B., & Liese, A. D. (2010) · Spatial and Spatio-temporal Epidemiology · doi:10.1016/j.sste.2010.09.008 · Cited by 7
“[PubMed: 12185896] Dabelea D, Bell RA, D’Agostino RB Jr, Imperatore G, Johansen JM, Linder B, et al. Incidence of diabetes in youth in the United States.”
Harrison, J. M., Breeze, M. L., Berman, K. H., & Harrigan, G. G. (2012) · Regulatory Toxicology and Pharmacology · doi:10.1016/j.yrtph.2012.12.002 · Cited by 6
“Gelman (2006) recommended a uniform prior distribution over a wide range, with zero as the minimum of that range, for the standard deviation of variance components in a hierarchical model when a proper prior distribution with low information is desired, and Kéry (2010) provided examples of WinBUGS…”
BackgroundGrilli, L., & Innocenti, F. (2017) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2017.1341886 · Cited by 6
“known that they strongly affect the performance of the estimation methods and the importance of the prior distribution [3, 11].”
BackgroundHydrographic responses to regional covariates across the Kara Sea
Mäkinen, J., & Vanhatalo, J. P. (2016) · Journal of Geophysical Research Oceans · doi:10.1002/2016jc011981 · Cited by 6
“We gave weakly informative half Student-t priors [Gelman, 2006] for all length-scale parameters and priors for the variance parameters of covariance functions.”
MethodQuantile POD for nondestructive evaluation with hit--miss data
Koh, Y.-M., & Meeker, W. Q. (2017) · Research in Nondestructive Evaluation · doi:10.1080/09349847.2017.1374493 · Cited by 6
“For σγ and σρ (non-negative parameters), following the recommendation of Gelman (2006), we used a Uniform(0, 100) prior distribution.”
MethodExplaining variability in metered energy use for similar buildings using Bayesian inference
Kristensen, M. H., & Petersen, S. (2017) · Energy Procedia · doi:10.1016/j.egypro.2017.09.709 · Cited by 6
Value assignment and uncertainty evaluation for single-element reference solutions
Possolo, A. M., Bodnar, O., Butler, T. A., Molloy, J. L., & Winchester, M. R. (2018) · Metrologia · doi:10.1088/1681-7575/aabd57 · Cited by 6
“We complete the specification of the model as a Bayesian statistical model: this involves using probability distributions to express states of knowledge about quantities that are not observable directly (which play the role of parameters in the Metrologia 55 (2018) 404 A Possolo et al model), and to link the measured…”
MethodInfluentialLangseth, B. J., Syslo, J. M., Yau, A., Kapur, M. S., & Brodziak, J. (2018) · NOAA Institutional Repository · doi:10.7289/v5/tm-pifsc-69 · Cited by 6
Heck, D. W. (2018) · Journal of Mathematical Psychology · doi:10.1016/j.jmp.2018.11.002 · Cited by 5
Held, F., Ekstrand, C., Cvijović, M., et al. (2019) · Journal of Pharmacokinetics and Pharmacodynamics · doi:10.1007/s10928-018-09617-0 · Cited by 6
Bayesian Approaches to Assessing the Parallel Lines Assumption in Cumulative Ordered Logit Models
Xu, J., Bauldry, S., & Fullerton, A. S. (2019) · Sociological Methods & Research · doi:10.1177/0049124119882461 · Cited by 7
“Following Gelman (2006), we use the halfCauchy distribution as priors for the scale hyperparameters of slopes in the model (see equation [16]), and results are presented in Table 5.2 The hyperparameter means of slopes from the hyperparameter model are similar to those from both the MLE and the…”
BackgroundMethodInfluentialJang, E. J., Nandram, B., Ko, Y., & Kim, D. H. (2020) · Statistics in Medicine · doi:10.1002/sim.8493 · Cited by 6
ten Hove, D., Jorgensen, T. D., & van der Ark, L. A. (2020) · Springer proceedings in mathematics & statistics · doi:10.1007/978-3-030-43469-4_7 · Cited by 9
“The choice among hyperprior distributions for random-effect variances is frequently discussed (see e.g., Gelman 2006; Gelman et al. 2013; Smid et al. 2019; Van Erp et al. 2019).”
BackgroundMethodResultInfluentialNavadeh, S., Mirzazadeh, A., McFarland, W., Coffin, P. O., Chehrazi, M., Mohammad, K., Nazemipour, M., Mansournia, M. A., McCandless, L. C., & Page, K. A. (2020) · Archives of Iranian Medicine · doi:10.34172/aim.2020.113 · Cited by 6
Variational Learning Data Fusion With Unknown Correlation
Zhang, W., Liang, Y., Leung, H., & Yang, F. (2022) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2021.3049769 · Cited by 6
Effectiveness of joint species distribution models in the presence of imperfect detection
Hogg, S., Wang, Y., & Stone, L. (2021) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.13614 · Cited by 6
Zoche-Golob, V., Pützschel, R., Einax, E., & Donat, K. (2021) · Irish Veterinary Journal · doi:10.1186/s13620-021-00204-3 · Cited by 6
A zero-inflated mixture spatially varying coefficient modeling of cholera incidences
Osei, F., Stein, A., & Andreo, V. (2022) · Spatial Statistics · doi:10.1016/j.spasta.2022.100635 · Cited by 6
“An alternative, suggested by Gelman (2006), is to assign a trictly positive half-normal (e.g. σku ∼ N+∞(010)) or uniform (e.g. σku ∼ U(010)) densities to their standard deviations.”
BackgroundMassey, P., Kearney, M., Rideau, A., Peterson, A., Gipson, J., Nianogo, R., Bornstein, M., Prelip, M., & Glik, D. (2022) · Journal of Global Health · doi:10.7189/jogh.12.04039 · Cited by 6
Kachel, S. M., Anderson, K. M., & Shokirov, Q. (2022) · Biological Conservation · doi:10.1016/j.biocon.2022.109588 · Cited by 7
Acevedo, J., Acevedo, M. L., Gaete-Argel, A., Rafael Araos, B., Gonzalez, C., Espinoza, D., Rivas, S., Pizarro, P., Jarpa, S., Soto‐Rifo, R., et al. (2022) · Clinical Microbiology and Infection · doi:10.1016/j.cmi.2022.11.017 · Cited by 6
“Following Gelman [12], independent uniform prior distributions were considered for the standard deviation of the random intercepts and errors.”
Breen, C. F., Mahmud, A. S., & Feehan, D. M. (2022) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1010742 · Cited by 6
Bayesian Spatial Modelling of HIV Prevalence in Jimma Zone, Ethiopia
Debusho, L. K., & Bedaso, N. G. (2023) · Diseases · doi:10.3390/diseases11010046 · Cited by 9
“In this study, non-informative priors were chosen for regression coefficients, or variances of random effects, because there was no empirical data on these model parameters [31].”
MethodA Bayesian hierarchical approach to the joint modelling of Revealed and stated choices
Li, Z., Washington, S. P., Zheng, Z., et al. (2023) · Journal of Choice Modelling · doi:10.1016/j.jocm.2023.100419 · Cited by 6
Sorel, M. H., Murdoch, A. R., Zabel, R. W., Kamphaus, C. M., Buhle, E. R., Scheuerell, M. D., & Converse, S. J. (2023) · Ecology and Evolution · doi:10.1002/ece3.10087 · Cited by 10
“Because there were only three streams available to estimate the hyperdistribution standard deviations h , we applied exponential regularizing penalties [e.g., h ∼ exp ( )] (Simpson et al., 2017), a common practice when fitting random- effect distributions with few levels (Gelman, 2006).”
MethodComparing estimation approaches for generalized additive mixed models with binary outcomes
Mullah, M. A. S., Hossain, Z., & Benedetti, A. (2023) · Journal of Statistical Computation and Simulation · doi:10.1080/00949655.2023.2222864 · Cited by 6
Cognitive and executive impairments in Parkinson’s disease psychosis: a Bayesian meta-analysis
Pisani, S., Gosse, L., Wieretilo, R., Ffytche, D. H., Velayudhan, L., & Bhattacharyya, S. (2023) · Journal of Neurology Neurosurgery & Psychiatry · doi:10.1136/jnnp-2022-331028 · Cited by 9
Evidence of seasonal heat acclimatisation in recreationally active adults during a mild summer
Brown, H. A., Topham, T. H., Clark, B., Flouris, A. D., Ioannou, L. G., Telford, R. D., Smallcombe, J. W., Jay, O., & Périard, J. D. (2023) · Journal of science and medicine in sport · doi:10.1016/j.jsams.2023.10.006 · Cited by 10
Combining reproducibility and repeatability studies with applications in forensic science
Arora, H., Kaplan‐Damary, N., & Stern, H. S. (2023) · Law Probability and Risk · doi:10.1093/lpr/mgad007 · Cited by 7
Chu, H., Lin, L., Wang, Z., Wang, Z., Chen, Y., & Cappelleri, J. C. (2023) · Wiley Interdisciplinary Reviews Computational Statistics · doi:10.1002/wics.1639 · Cited by 7
“For our AB-NMA models, we adapted the source code from two R functions in the pcnetmeta package (Lin, Zhang, et al., 2017), and used vague priors N(0, 10 4 ) for the treatment-specific fixed effects, Unif(0.0001, 10) for the standard deviation of the random effects (Gelman, 2006).”
MethodInfluential
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.