"Everything should be made as simple as possible, but not simpler." - Albert Einstein

Seminar Announcement

A Survey Analysis of the 2006 National Survey of Fishing, Hunting, and Wildlife- Associated Recreation Wildlife- Watching Activities

Ryan Koralik, M.S. Candidate, Department of Statistics, Colorado State University

Monday, July 12, 2010

3:00 p.m., room 006, Statistics Bldg

ABSTRACT

The National Survey of Fishing, Hunting, and Wildlife- Associated Recreation (FHWAR) is conducted by the U.S. Census Bureau to identify participation trends in different wildlife- associated activities and the trends of state and national park usage for such activities. This paper considered model- based and design- based methods to analyze the 2006 version of the dataset. The model based methods used were the ordinary least squares and logistic regression methods. The design- based methods used were the design- weighted method and the design- weighted method, also using replicate weights. The Pfeffermann- Sverchkov semi- parametric method was also considered when the design was found to be informative. All of the methods were compared to each other and their selected best models were also compared in the case study. Due to a large number of variables considered in the study, model selection criteria were used. This paper discusses the concept of model selection in the survey design- weighted context. Since design weights were given, survey design- weighted versions of three frequently used model selection statistics were created and compared to their ordinary versions. The goal of this project was to compare different design- weighted methods to the frequently used ordinary model- based methods and to take properly take the design into account, especially in scenarios where the design was found to be informative.

Advisory Committee:

Dr. Jean Opsomer, Advisor
Dr. Jay Breidt, Committee Member
Dr. Jerry Vaske, Human Dimensions of Natural Resources, Outside Member