Residual analysis for discrete correlated data in the multivariate approach
Main Article Content
Abstract
The residual distributions obtained from discrete correlated and uncorrelated data cannot be well approximated to the standardized normal distribution. In this case, the efficiency in checking the adequacy of the model to the data and detecting outliers is not guaranteed. Thus, alternative measures for residual analysis have been considered in several classes of models and their properties have been assessed. In this paper, we investigate the empirical distribution of four residuals of the multivariate negative binomial regression (MNBR) model. In our study, we propose standardized weighted and standardized Pearson residuals; we also consider the standardized component of deviance and quantile residuals suggested by Fabio et al. (2012) and Fabio et al. (2023), respectively. Monte Carlo simulation results reveal that the concordance of the empirical distribution of the residuals to the standard normal distribution depends on the dispersion parameter. Furthermore, the impact on residual analysis when the random effect distribution is misspecified is explored. We concluded that the quantile and standardized weighted residuals presented better performances.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
References
Diggle, P. J., Liang, K. Y. & Zeger, S. L. Analysis of Longitudinal Data 2nd ed. (Oxford University Press, N.Y., 2013). https://doi.org/10.1093/oso/9780198524847.001.0001
Espinheira, P. L., S., F. & Cribari-Neto, F. On beta regression residuals. Journal of Applied Statistics 35, 407–419 (2008). https://doi.org/10.1080/02664760701834931
Fabio, L. C., Paula, G. A. & de Castro, M. A Poisson mixed model with nonormal random effect distribution. Computational Statistics and Data Analysis 56, 1499–1510 (2012). https://doi.org/10.1016/j.csda.2011.12.002
Fabio, L. C., Villegas, C., Carrasco, J. M. F. & de Castro, M. Diagnostic tools for a multivariate negative binomial model for fitting correlated data with overdispersion. Communications in Statistics - Theory and Methods. 52, 1833–1853 (2023). https://doi.org/10.1080/03610926.2021.1939380
Feng C., L., L. & Sadeghpour, A. A comparison of residual diagnosis tools for diagnosing regression models for count data. BMC Medical Research Methodology 20, 1–21 (2020). https://doi.org/10.1186/s12874-020-01055-2
Hand, D. J. & Crowder, M. Practical Longitudinal Data Analysis (London: Chapman & Hall, 1996). https://doi.org/10.1201/9780203742372
Hand, D. J. & Taylor, C. C. Analysis of Variance and Repeated Measures (London: Chapman & Hall, 1987).
Hardin, J. W. & Hilbe, J. M. Generalized Linear Models and Extensions, Second Edition (Stata Press, Texas, 2016).
Johnson, N., Kotz, S. & Balakrishnan, N. Discrete Multivariate Distributions (Wiley, New York, 1997). https://doi.org/10.2307/3109790
Lawless, J. Negative binomial and mixed Poisson regression. The Canadian Journal of Statistics 15, 209–225 (1987). https://doi.org/10.2307/3314912
Pereira, G. H. A., Scudilio, J., Santos-Neto, M., Botter, D. A. & Sandoval, M. C. A class of residuals for outlier identification in zero adjusted regression models. Journal of Applied Statistics 47, 1833–1847 (2020). https://doi.org/10.1080/02664763.2019.1696759
Scudilio, J. & Pereira, G. H. A. Adjusted quantile residual for generalized linear models. Computational Statistics 35, 399–421 (2020). https://doi.org/10.1007/s00180-019-00896-w
Tsui, K.-W. Multiparameter estimation for some multivariate discrete distributions with possibly dependent components. Annals of the Institute of Statistical Mathematics 38, 45–56 (1986). https://doi.org/10.1007/BF02482499
Waller, L. A. & Zelterman, D. Log-Linear Modeling with the Negative Multinomial Distribution. Biometrics 53, 971–982 (1997). http://dx.doi.org/10.2307/2533557