|
Activity Number:
|
362
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics & the Environment
|
| Abstract - #302076 |
|
Title:
|
Exploring Design and Model-Based Approaches for Handling Nonignorable Missing Data Mechanism for Inference in Environmental Surveys
|
|
Author(s):
|
Breda Munoz-Hernandez*+ and Virginia M. Lesser and Ruben A. Smith
|
|
Companies:
|
Oregon State University/Universidad De Costa Rica and Oregon State University and Oregon State University
|
|
Address:
|
Escuela De Matematica, San Jose, , , Costa Rica
|
|
Keywords:
|
nonignorable ; missing data ; response probability ; environmental surveys
|
|
Abstract:
|
Missing data occurs in environmental surveys for many reasons, and one of them is failing to access the sampling site. When the owners of the land where the sampling site is located deny the access to the site, one may suspect a possible association between the nongranted access and the environmental variable being measure. Situations like this are results of what is known in the survey literature as a nonignorable missing mechanism. To handle this type of missing data, we consider in this paper a nonresponse model for nonignorable missing mechanism and assume that the covariates incorporated in the modeling of the nonresponse model are fully observed. The approach centers in the formulation of the likelihood function that incorporates the response probability and posterior estimation of the parameters involved. The theoretical development of the methodology is discussed and an illustration with data from environmental surveys is also presented.
|
