Al-Bahir Journal for Engineering and Pure Sciences


Bayesian inferences depend solely on specification and accuracy of likelihoods and prior distributions of the observed data. The research delved into Bayesian estimation method of regression models to reduce the impact of some of the problems, posed by convectional method of estimating regression models, such as handling complex models, availability of small sample sizes and inclusion of background information in the estimation procedure. Posterior distributions are based on prior distributions and the data accuracy, which is the fundamental principles of Bayesian statistics to produce accurate final model estimates. Sensitivity analysis is an essential part of mathematical model validation in obtaining a robust inference. Prior sensitivity analysis was examined in regression model, via Bayesian regression and Bayesian quantile regression analysis; results obtained across the sensitivity analysis were compared using RMSE and BIAS statistic as model performance evaluation. Empirical studies using Nigeria Economic variables were employed to analyze the variation in prior sensitivity. Different hyper parameters of the priors were used to check for sensitivity of the prior distributions, it was ascertained that Bayesian method under the frame work of regression quantiles performs well with small variance and sample size than Bayesian regression methods.



[1] Ayoade Adewole, Oiusoga Fasoranbaku. Bayesian Estimation of Regression quantiles in the presence of Autocorrelated Errors. Tanzan J Sci 2020;46(1):76.

[2] Hoeting Jennifer A, Madigan David, Raftery Adrian E, Volinsky Chris T. Bayesian model averaging: a tutorial with comments by M. clyde, David Draper and E.I. George and a rejoinder by the authors. Statist. Sci. 1999;14(4):382e417.

[3] Clyde M, George EI. Model Uncertainty. Stat Sci 2004;19(1): 81e94.

[4] Gelman A, et al. A weakly informative default prior distribution for logistic and other regression models. Ann Appl Stat 2008;2:1360e83.

[5] Zondervan-Zwijnenburg MAJ, Peeters M, Depaoli S, van de Schoot R. Where do priors come from? Applying guidelines to construct informative priors in small sample research. Res Hum Dev 2017;14:305e20.

[6] Zondervan-Zwijnenburg MAJ, Depaoli S, Peeters M, van de Schoot R. Pushing the limits: the performance of ML and Bayesian estimation with small an unbalanced samples in a latent growth model. Methodology 2019;15:31e43.

[7] Depaoil S, Cliffon JP. A Bayesian approach to multilevel Structural equation modelling with continuous and dichoomous outcomes. Struct Equ Model 2015;22(3):327e51.

[8] Kruschke John K. Bayesian Estimation Supersedes the t Test. J Exp Psychol 142(2), 573-603.

[9] Adewole Ayoade I, Loyinmi AC. Residual Bootstraps for Regression Model Validation. Sci World J 2019;14(2): 1597e6343.

[10] Van de Schoot R, Sijbrandij M, Depaoli S, Winter S, Olff M, van Loey N. Bayesian PTSD-trajectory analysis with informed priors based on a systematic literature search and expert elicitation. Multiv. Behav. Res. 2018;53:267e91.

[11] Edwards Kelly D, Konoids Timothy R. Impact of Informative priors on Model Fit indices in Bayesian Confirmatory Factor Analysis, Structural Equation Modelling. A Multidiscip J 2022. https://doi.org/10.1080/10705511.2022.212635.

[12] Grzenda Wioletta. Informative and Non informative Prior Distributions and their impact on the accuracy of Bayesian Inference. Statistics in Transition 2016;17(4):763e80.

[13] Depaoil Sonja D Winter, MariekeVisser. The importance of Prior Sensitivity Analysis in Bayesian Statistics: Demonstrations using an interactive Shinny App. Frontier Psychology 2020;11:1e47.

[14] Vanpaemel W. Prior sensitivity in theory testing: An apologia for the Bayes factor. J Math Psychol 2020:1e8. https://doi.org/10.1016/j.jmp.2010.07.003. AL-BAHIR JOURNAL FOR ENGINEERING AND PURE SCIENCES 2024;4:88e96 95

[15] Joshi Chaitanya, Ruggeri Fabrizio. On a class of prior distributions that accounts for uncertainty in the data. Int J Approx Reason 2023;161:1e19.

[16] Kallioinen Noa, Paananen Topi, Bürkner Paul-Christian, Aki Vehtari. Detecting and diagnosing prior and likelihoosensitivity with power-scaling. 2023. arXiv:2107.14054v4 [ stat.ME] 1-23.

[17] Babanezhad Manoochehr. Bayesian and Non-Bayesian Sensitivity Analysis for Correcting Measurement Error. J Appl Math Bioinf 2012;2(1):21e7.

[18] Seong Mi Choi and Dal Ho Kim. Sensitivity analysis in Bayesian nonignorable selection model for binary responses. Journal of the Korean Data 2014;25(1):187e94.

[19] Roos Ma!lgorzata, Held Leonhard. Sensitivity analysis in Bayesian generalized linear mixed models for binary data. Bayesian Analysis 2011;6(2):259e78.

[20] da Paz Rosineide F, Balakrishnan Narayanaswamy, Luis Bazan Jorge. L-Logistic regression models: Prior sensitivity analysis, robustness to outliers and applications Brazilian. Journal of Probability and Statistics 2018;1:1e24.

[21] Berger J. An overview of robust Bayesian analysis. Test 1994; 3(1):5e124.

[22] Berger JO, Insua DR, Ruggeri F. Bayesian Robustness. In: Insua DR, Ruggeri F, editors. Robust Bayesian analysis. New York: Springer; 2000. p. 1e32.

[23] Watson J, Holmes C. Approximate Models and Robust Decisions. Stat Sci 2016;31(4):465e89.

[24] Van de schoot, Herbert Hoijtink, Romoijn Jan- Willem, Brugman Daniel. A prior Predictive loss functions the evaluation of inequality constrained hypothesis. J Math Psychol 2012;56(1):13e23.

[25] Koenker R, Basset G. Regression quantiles. Econometrica 1978;46:33e50.

[26] Yu K, Moyeed RA. Bayesian Quantile Regression. Stat Probab Lett 2001;54(4):437e47.

[27] Liu Hong G. Simulating sensitivities of Conditional value at Risk. Manag Sci 2008;55(2):281e93.

[28] Hong LJ. Estimating Quantile Sensitivities. Oper Res 2009;57: 118e30.

[29] Volk- Makarewicz Waren, Heidergott. Sensitivity Analysis of Quantiles. AENORM 2012;20(75):26e31.

[30] Kozumi H, Kobayash G. Gibbs sampling methods for Bayesian quantile regression. J Stat Comput Simulat 2011;81: 1565e78.