Abstract
There has been extensive development of diagnostic measures for Generalized Linear Models fitted by the maximum likelihood method. However, there is evidence that the maximum likelihood estimator is extremely sensitive to outlying, leverage and influential observations. We propose a diagnostic measure based on minimum distance estimates to assess the effect that the estimation method has on parameter estimates. Furthermore, a new single case deletion diagnostic to detect leverage observations is developed. Finally, the paper concludes with an analysis of real data.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bedrick EJ, Tsai C (1993) Diagnostics for binomial response models using power divergence statistics. Comput Stat Data Anal 15:381–392
Cook RD (1977) Detection of influential observations in linear regression. Technometrics 19:15–18
Cressie NAC, Read T (1984) Multinomial goodness-of-fit tests. J R Stat Soc Ser B 46:440–464
Fahrmeir L, Tutz G (2001) Multivariate statistical modelling based on generalized linear models. Springer, New York
Martin N, Pardo L (2010) A new measure of leverage cells in multinomial loglinear models. Commun Stat-Theor Meth 39:517–530
Nelder JA, Wedderburn RWM (1972) Generalized linear models. J R Stat Soc Ser A 135:370–384
Pardo L (2006) Statistical inference based on divergence measures. Chapman & Hall, London
Pardo MC (2010) High Leverage points and outliers in generalized linear models for ordinal data. In: Skiadas CH (ed) Advances in data analysis, statistics for industry and technology. Birkhäuser, Boston
Pardo MC (2011) Testing equality restrictions in generalized linear models. Metrika 73(2):231–253
Pardo MC, Alonso R (2012) Influence measures based on the volume of confidence ellipsoids for GEE. Biom J 54(4):552–567
Pardo JA, Pardo MC (2008) Minimum phi-divergence estimator and phi-divergence statistics in generalized linear models with binary data. Methodol Comput Appl Probab 10(3):357–379
Pardo JA, Pardo MC (2009) Selecting and checking a binomial response generalized linear model using phi-divergences. J Stat Plan Inference 139:3038–3049
Pardo MC, Pardo JA (2011) Detection of outlying points in ordered polytomous. In: Pardo L, Balakrishnan N, Gil MA (eds) Modern mathematical tools and techniques in capturing complexity. Springer, Berlin
Pardo JA, Pardo L, Pardo MC (2006) Minimum $\phi $-divergence estimator in logistic regression models. Stat Pap 47:91–108
Parr WC (1981) Minimum distance estimation: a bibliography. Commun Stat-Theor Meth 12:1205–1224
Vajda I (1989) Theory of statistical inference and information. Kluwer Academic Publishers, Dordrecht
Wolfowitz J (1953) Estimation by the minimum distance method. Ann Inst Stat Math 5:9–23
Acknowledgements
This work was partially supported by Grant MTM2013-40778-R. To the memory of Dr. Pedro Gil, an excellent statistician and great person.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2018 Springer International Publishing AG
About this chapter
Cite this chapter
Pardo, M.C. (2018). Tools for Detecting Leverage and Influential Observations in Generalized Linear Models. In: Gil, E., Gil, E., Gil, J., Gil, M. (eds) The Mathematics of the Uncertain. Studies in Systems, Decision and Control, vol 142. Springer, Cham. https://doi.org/10.1007/978-3-319-73848-2_47
Download citation
DOI: https://doi.org/10.1007/978-3-319-73848-2_47
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-73847-5
Online ISBN: 978-3-319-73848-2
eBook Packages: EngineeringEngineering (R0)