Dissertation/Thesis Abstract

Generalized linear regression models for count data
by Nava, Martha, M.S., California State University, Long Beach, 2014, 54; 1528005
Abstract (Summary)

Count data are observations of only non-negative integer values (0, 1, 2, etc.). When the response variable follows a Poisson distribution, the Poisson regression may be used to model the data. The main feature of a Poisson distribution is that the mean is equal to the variance. If this condition does not hold, and the variance is much larger than the mean, overdispersion occurs. When this arises, a negative binomial regression model may be employed. If there are more zeros observed than normal for a Poisson (or negative binomial) regression, a zero-inflated Poisson (or negative binomial) regression model may be applicable. If zeros are not observed due to the data not containing zero counts, then a zero-truncated Poisson (or zero-truncated negative binomial) model may be used.

This research will explain these models with examples as well as the test for overdispersion using SAS and R software.

Indexing (document details)
Advisor: Korosteleva, Olga
Commitee:
School: California State University, Long Beach
School Location: United States -- California
Source: MAI 53/01M(E), Masters Abstracts International
Source Type: DISSERTATION
Subjects: Statistics
Keywords: Count data, Hurdle, Over dispersion, Poisson, Zero inflated, Zero truncated
Publication Number: 1528005
ISBN: 978-1-303-98529-4
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy
ProQuest