Abstract
Zero-adjusted generalized linear models (ZAGLMs) are used in many areas to fit variables that are discrete at zero and continuous on the positive real numbers. As in other classes of regression models, hypothesis testing inference in the class of ZAGLMs is usually performed using the likelihood ratio statistic. However, the LR test is substantially size distorted when the sample size is small. In this work, we derive an analytical Bartlett correction of the LR statistic. We also consider two different adjustments for the LR statistic based on bootstrap. Monte Carlo simulation studies show that the improved LR tests have null rejection rates close to the nominal levels in small sample sizes and similar power. An application illustrates the usefulness of the improved statistics.
Similar content being viewed by others
References
Araújo MC, Cysneiros AH, Montenegro LC (2020) Improved heteroskedasticity likelihood ratio tests in symmetric nonlinear regression models. Stat Pap 61:167–188
Bartlett MS (1937) Properties of sufficiency and statistical tests. Proceedings of the royal society of London A: Mathematical, Physical and Engineering Sciences 160(901):268–282
Bayer FM, Cribari-Neto F (2013) Bartlett corrections in beta regression models. J Stat Plan Inference 143(3):531–547
Bortoluzzo AB, Claro DP, Caetano MAL, Artes R (2011) Estimating total claim size in the auto insurance industry: a comparison between tweedie and zero-adjusted inverse gaussian distribution. Braz Adm Rev 8(1):37–47
Botter DA, Cordeiro GM (1997) Bartlett corrections for generalized linear models with dispersion covariates. Commun Stat Theory Methods 26(2):279–307
Calsavara VF, Rodrigues AS, Rocha R, Louzada F, Tomazella V, Souza AC, Costa RA, Francisco RPV (2019) Zero-adjusted defective regression models for modeling lifetime data. J Appl Stat 46(13):2434–2459
Cordeiro GM (1983) Improved likelihood ratio statistics for generalized linear models. J R Stat Soc B 45(3):404–413
Cordeiro GM, Cribari-Neto F (2014) An introduction to Bartlett correction and bias reduction. Springer, Berlin
Cordeiro GM, Paula GA, Botter DA (1994) Improved likelihood ratio tests for dispersion models. Int Stat Rev 62(2):257–274
Cox DR, Reid N (1987) Parameter orthogonality and approximate conditional inference. J R S Soc B 49(1):1–39
Das U, Dhar SS, Pradhan V (2018) Corrected likelihood-ratio tests in logistic regression using small-sample data. Commun Stat Theory Methods 47(17):4272–4285
Doornik JA (2009) An object-oriented matrix language: Ox 6. Timberlake Consultants Press, London
Dunn PK, Smyth GK (2018) Generalized linear models with examples in R. Springer, Berlin
Efron B et al (1979) Bootstrap methods: another look at the jackknife. Ann Stat 7(1):1–26
Guedes AC, Cribari-Neto F, Espinheira PL (2020) Modified likelihood ratio tests for unit gamma regressions. J Appl Stat 47(9):1562–1586
Guedes AC, Cribari-Neto F, Espinheira PL (2021) Bartlett-corrected tests for varying precision beta regressions with application to environmental biometrics. PLoS ONE 16(6):e0253349
Hashimoto EM, Ortega EM, Cordeiro GM, Cancho VG, Klauberg C (2019) Zero-spiked regression models generated by gamma random variables with application in the resin oil production. J Stat Comput Simul 89(1):52–70
Heller G, Stasinopoulos M, Rigby B The zero-adjusted inverse gaussian distribution as a model for insurance claims, in: J. Hinde, J. Einbeck, J. Newell (Eds.), Proceedings of the 21th International Workshop on Statistical Modelling, Ireland, Galway, 2006, pp. 226–233
Lambert D (1992) Zero-inflated poisson regression, with an application to defects in manufacturing. Technometrics 34(1):1–14
Lawley DN (1956) A general method for approximating to the distribution of likelihood ratio criteria. Biometrika 43(3/4):295–303
Loose LH, Bayer FM, Pereira TL (2017) Bootstrap bartlett correction in inflated beta regression. Commun Stat-Simul Comput 46(4):2865–2879
Loose LH, Valença DM, Bayer FM (2018) On bootstrap testing inference in cure rate models. J Stat Comput Simul 88(17):3437–3454
Magalhães TM, Gallardo DI (2020) Bartlett and bartlett-type corrections for censored data from a weibull distribution. Stat Op Res Trans 44(1):127–140
Melo TF, Ferrari SL, Cribari-Neto F (2009) Improved testing inference in mixed linear models. Comput Stat Data Anal 53(7):2573–2582
Melo TF, Vargas TM, Lemonte AJ, Moreno-Arenas G (2022) Higher-order asymptotic refinements in the multivariate dirichlet regression model. Commun Stat-Simul Comput 51(1):53–71
Michaelis P, Klein N, Kneib T (2020) Mixed discrete-continuous regression-a novel approach based on weight functions. Stat 9(1):e277
Moulton LH, Weissfeld LA, Laurent RTS (1993) Bartlett correction factors in logistic regression models. Comput Stat Data Anal 15(1):1–11
Nogarotto DC, Azevedo CLN, Bazán JL et al (2020) Bayesian modeling and prior sensitivity analysis for zero-one augmented beta regression models with an application to psychometric data. Braz J Probab Stat 34(2):304–322
Ospina R, Ferrari S (2012) A general class of zero-or-one inflated beta regression models. Comput Stat Data Anal 56(6):1609–1623
Pereira TL, Cribari-Neto F (2014) Modified likelihood ratio statistics for inflated beta regressions. J Stat Comput Simul 84(5):982–998
Pereira GH, Scudilio J, Santos-Neto M, Botter DA, Sandoval MC (2020) A class of residuals for outlier identification in zero adjusted regression models. J Appl Stat 47(10):1833–1847
Rauber C, Cribari-Neto F, Bayer FM (2020) Improved testing inferences for beta regressions with parametric mean link function. Adv Stat Anal 104(4):687–717
Rocha LO, Nakai VK, Braghini R, Reis TA, Kobashigawa E, Corrêa B (2009) Mycoflora and co-occurrence of fumonisins and aflatoxins in freshly harvested corn in different regions of brazil. Int J Mol Sci 10(11):5090–5103
Rocha LO, Reis GM, Fontes LC, Piacentini KC, Barroso VM, Reis TA, Pereira AA, Corrêa B (2017) Association between fum expression and fumonisin contamination in maize from silking to harvest. Crop Prot 94:77–82
Rocke DM (1989) Bootstrap bartlett adjustment in seemingly unrelated regression. J Am Stat Assoc 84(406):598–601
Rubec PJ, Kiltie R, Leone E, Flamm RO, McEachron L, Santi C (2016) Using delta-generalized additive models to predict spatial distributions and population abundance of juvenile pink shrimp in tampa bay, florida. Marine Coastal Fish 8(1):232–243
Sen PN, Singer JM, Lima ACP (2010) From finite sample to asymptotic methods in statistics. Cambridge University Press, Cambridge
Silva AR, Azevedo CL, Bazán JL, Nobre JS (2021) Augmented-limited regression models with an application to the study of the risk perceived using continuous scales. J Appl Stat 48(11):1998–2021
Skovgaard IM (2001) Likelihood asymptotics. Scand J Stat 28(1):3–32
Thomson FJ, Letten AD, Tamme R, Edwards W, Moles AT (2018) Can dispersal investment explain why tall plant species achieve longer dispersal distances than short plant species? New Phytol 217(1):407–415
Tomazella V, Nobre JS, Pereira GH, Santos-Neto M (2019) Zero-adjusted birnbaum-saunders regression model. Statist Probab Lett 149:142–145
Tong E, Mues C, Thomas L (2013) A zero-adjusted gamma model for mortgage loan loss given default. Int J Forecast 29:548–562
Tong EN, Mues C, Brown I, Thomas LC (2016) Exposure at default models with and without the credit conversion factor. Eur J Op Res 252(3):910–920
Ye T, Lachos VH, Wang X, Dey DK (2021) Comparisons of zero-augmented continuous regression models from a Bayesian perspective. Stat Med 40(5):1073–1100
Zamani H, Bazrafshan O (2020) Modeling monthly rainfall data using zero-adjusted models in the semi-arid, arid and extra-arid regions. Meteorol Atmos Phys 132(2):239–253
Acknowledgements
We acknowledge the two referees’ suggestions, which helped us to improve this work.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
The remaining quantities to define the Bartlett correction factor, see equation (8), are:
\({{\varvec{F}}} = \text{ diag }\{f_{11}, \ldots , f_{nn}\}\), \({{\varvec{G}}} = \text{ diag }\{g_{11}, \ldots , g_{nn}\}\), \({{\varvec{F}}}_{\pi } = \text{ diag }\{f_{\pi 11}, \ldots , f_{\pi nn}\}\), \({{\varvec{G}}}_{\pi } = \text{ diag }\{g_{\pi 11}, \ldots , g_{\pi nn}\}\), \({{\varvec{H}}}_{i} = \text{ diag }\{h_{i,11}, \ldots , h_{i,nn}\}\), \(i = 1, 2, 3\), where
Although deriving the expression for \(\varepsilon _p\) entails a great deal of algebra, this expression only involves simple operations of diagonal matrices, i.e. they are simple expressions to implement.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Magalhães, T.M., Pereira, G.H.A., Botter, D.A. et al. Bartlett corrections for zero-adjusted generalized linear models. Stat Papers (2023). https://doi.org/10.1007/s00362-023-01477-2
Received:
Revised:
Published:
DOI: https://doi.org/10.1007/s00362-023-01477-2