Inference in a generalized endpoint-inflated binomial regression model. Issue 4 (4th July 2017)

• Record Type: Journal Article
• Authors: Dupuy, Jean-François
• Is Part Of: Statistics. Volume 51:Issue 4(2017)
• Page Start: 888
• Page End: 903
• Publication Date: 2017-07-04
• Subjects: Count data -- large-sample properties -- logistic regression -- simulations; Mathematical statistics -- Periodicals; 519.505
• DOI: 10.1080/02331888.2017.1316724
• Languages: English
• ISSNs: 0233-1888

Abstract

ABSTRACT: Generalized endpoint-inflated binomial regression was recently proposed to model count data with large frequencies of both zeros and right-endpoints. Maximum likelihood estimation (MLE) was developed for this model and simulations suggest that the resulting estimates behave well. However, large-sample properties of the MLE have not yet been rigorously established. Such results are however essential for ensuring reliable statistical inference and decision-making. This paper addresses this issue. Identifiability of the generalized endpoint-inflated binomial regression model is first proved. Then, consistency and asymptotic normality of the MLE are established. A simulation study is conducted to assess finite-sample behaviour of the estimator.