##
**A Bayesian formulation of exploratory data analysis and goodness-of-fit testing.**
*(English)*
Zbl 1114.62320

Summary: Exploratory data analysis (EDA) and Bayesian inference (or, more generally, complex statistical modeling)—which are generally considered as unrelated statistical paradigms—can be particularly effective in combination. In this paper, we present a Bayesian framework for EDA based on posterior predictive checks. We explain how posterior predictive simulations can be used to create reference distributions for EDA graphs, and how this approach resolves some theoretical problems in Bayesian data analysis. We show how the generalization of Bayesian inference to include replicated data \(y^{\text{rep}}\) and replicated parameters \(\theta^{\text{rep}}\) follows a long tradition of generalizations in Bayesian theory.

On the theoretical level, we present a predictive Bayesian formulation of goodness-of-fit testing, distinguishing between \(p\)-values (posterior probabilities that specified antisymmetric discrepancy measures will exceed 0) and \(u\)-values (data summaries with uniform sampling distributions). We explain that \(p\)-values, unlike \(u\)-values, are Bayesian probability statements in that they condition on observed data.

Having reviewed the general theoretical framework, we discuss the implications for statistical graphics and exploratory data analysis, with the goal being to unify exploratory data analysis with more formal statistical methods based on probability models. We interpret various graphical displays as posterior predictive checks and discuss how Bayesian inference can be used to determine reference distributions.

The goal of this work is not to downgrade descriptive statistics, or to suggest they be replaced by Bayesian modeling, but rather to suggest how exploratory data analysis fits into the probability-modeling paradigm.

We conclude with a discussion of the implications for practical Bayesian inference. In particular, we anticipate that Bayesian software can be generalized to draw simulations of replicated data and parameters from their posterior predictive distribution, and these can in turn be used to calibrate EDA graphs.

On the theoretical level, we present a predictive Bayesian formulation of goodness-of-fit testing, distinguishing between \(p\)-values (posterior probabilities that specified antisymmetric discrepancy measures will exceed 0) and \(u\)-values (data summaries with uniform sampling distributions). We explain that \(p\)-values, unlike \(u\)-values, are Bayesian probability statements in that they condition on observed data.

Having reviewed the general theoretical framework, we discuss the implications for statistical graphics and exploratory data analysis, with the goal being to unify exploratory data analysis with more formal statistical methods based on probability models. We interpret various graphical displays as posterior predictive checks and discuss how Bayesian inference can be used to determine reference distributions.

The goal of this work is not to downgrade descriptive statistics, or to suggest they be replaced by Bayesian modeling, but rather to suggest how exploratory data analysis fits into the probability-modeling paradigm.

We conclude with a discussion of the implications for practical Bayesian inference. In particular, we anticipate that Bayesian software can be generalized to draw simulations of replicated data and parameters from their posterior predictive distribution, and these can in turn be used to calibrate EDA graphs.

### MSC:

62F15 | Bayesian inference |

62G10 | Nonparametric hypothesis testing |

65C60 | Computational problems in statistics (MSC2010) |

### Keywords:

Bootstrap; Fisher’s exact test; Graphics; Mixture model; Model checking; Multiple imputation; Prior predictive check; Posterior predictive check; \(p\)-value; \(u\)-value### References:

[1] | Berger, Statistical Decision Theory and Bayesian Analysis (1985) |

[2] | Berkhof, Posterior predictive checking using antisymmetric discrepancy functions (2002) |

[3] | Box, Sampling and Bayes inference in scientific modelling and robustness., Journal of the Royal Statistical Society A 143 pp 383– (1980) · Zbl 0471.62036 |

[4] | Box, Bayesian Inference in Statistical Analysis (1973) |

[5] | Buja, Interactive high-dimensional data visualization., Journal of Computational and Graphical Statistics 5 pp 78– (1996) |

[6] | Buja , A. Cook , D. Swayne , D. 1999 Inference for data visualization Talk given at Joint Statistical Meetings |

[7] | Carlin , B.P. Banerjee , S. 2003 Hierarchical multivariate CAR models for spatio-temporally correlated survival data Bayesian Statistics 7, to appear |

[8] | Carlin, Bayes and Empirical Bayes Methods for Data Analysis (1996) · Zbl 0871.62012 |

[9] | Chambers, Graphical Methods for Data Analysis (1983) |

[10] | Cleveland, The Elements of Graphing Data (1985) |

[11] | Cleveland, Envisioning Information (1993) |

[12] | Clyde, Flexible empirical Bayes estimation for wavelets., Journal of the Royal Statistical Society B 62 pp 681– (2000) · Zbl 0957.62006 |

[13] | Denison, Bayesian Methods for Nonlinear Classification and Regression (2002) · Zbl 0994.62019 |

[14] | Draper, Assessment and propagation of model uncertainty (with discussion)., Journal of the Royal Statistical Society B 57 pp 45– (1995) · Zbl 0812.62001 |

[15] | Efron, Data analysis using Stein’s estimator and its generalizations., Journal of the American Statistical Association 70 pp 311– (1975) · Zbl 0319.62018 |

[16] | Efron, An Introduction to the Bootstrap (1993) |

[17] | Ehrenberg, Data Reduction: Analysing and Interpreting Statistical Data (1975) |

[18] | Gelman, Discussion of ”Analysis of non-randomly censored ordered categorical longitudinal data from analgesic trials” by L.B. Sheiner, S.L. Beal and A. Dunne., Journal of the American Statistical Association 92 pp 1248– (1997) |

[19] | Gelman, Bayesian Data Analysis (1995) |

[20] | Gelman, Diagnostic checks for discrete-data regression models using posterior predictive simulations., Applied Statistics 49 pp 247– (2000) · Zbl 0945.62071 |

[21] | Gelman, Posterior predictive assessment of model fitness via realized discrepancies (with discussion)., Statistica Sinica 6 pp 733– (1996) · Zbl 0859.62028 |

[22] | Gelman, Teaching Statistics: A Bag of Tricks. (2002) · Zbl 1020.97500 |

[23] | Gelman, Let’s practice what we preach: turning tables into graphs., The American Statistician 56 pp 121– (2002) |

[24] | Gelman, All maps of parameter estimates are misleading., Statistics in Medicine 18 pp 3221– (1999) |

[25] | Gelman, Bayesian model checking for missing and latent data problems using posterior predictive simulations (2002) |

[26] | Hill, Inference about variance components in the one-way model., Journal of the American Statistical Association 60 pp 806– (1965) · Zbl 0127.36004 |

[27] | Lindley, Bayes estimates for the linear model., Journal of the Royal Statistical Society B 34 pp 1– (1972) · Zbl 0246.62050 |

[28] | Louis, Estimating a population of parameter values using Bayes and empirical Bayes methods., Journal of the American Statistical Association 78 pp 393– (1984) |

[29] | Meulders, Assumptions, Robustness, and Estimation Methods in Multivariate Modeling (1998) |

[30] | Newton, Discovering combinations of genomic alterations associated with cancer., Journal of the American Statistical Association 97 pp 931– (2002) · Zbl 1041.62093 |

[31] | Raftery , A.E. 1995 Bayesian model selection in social research (with discussion) Sociological Methodology 1995 P.V. Marsden |

[32] | Ripley, Statistical Inference for Spatial Processes. (1988) · Zbl 0716.62100 |

[33] | Rubin, Inference and missing data., Biometrika 63 pp 581– (1976) · Zbl 0344.62034 |

[34] | Rubin, Estimation in parallel randomized experiments., Journal of Educational Statistics 6 pp 377– (1981) |

[35] | Spiegelhalter, Bayesian measures of model complexity and fit., Journal of the Royal Statistical Society B 64 pp 583– (2002) · Zbl 1067.62010 |

[36] | Spiegelhalter, BUGS: Bayesian inference using Gibbs sampling (1994) |

[37] | Stein, Inadmissibility of the usual estimator for the mean of a multivariate normal distribution pp 197– (1955) |

[38] | Tiao, Bayesian analysis of random-effect models in the analysis of variance. 1: Posterior distribution of variance components., Biometrika 52 pp 37– (1965) · Zbl 0144.42204 |

[39] | Tufte, The Visual Display of Quantitative Information (1983) |

[40] | Tufte, Envisioning Information (1990) |

[41] | Tukey, Statistical Papers in Honor of George W. Snedecor (1972) |

[42] | Tukey, Exploratory Data Analysis (1977) |

[43] | Wahba, Improper priors, spline smoothing and the problem of guarding against model errors in regression., Journal of the Royal Statistical Society B 40 pp 364– (1978) · Zbl 0407.62048 |

[44] | Wainer, Visual Revelations (1997) |

This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.