Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Master of Science

Program

Statistics and Actuarial Sciences

Supervisor

Bonner, Simon J.

Abstract

Mark-recapture studies are often used to estimate the survival of individuals in a population and identify factors that affect survival in order to understand how the population might be affected by changing conditions. Factors that vary between individuals and over time, like body mass, present a challenge because they can only be observed when an individual is captured. Several models have been proposed to deal with the missing-covariate problem and commonly impose a logit link function which implies that the survival probability varies between 0 and 1. In this thesis I explore the estimability of four possible models when survival is linked to the covariate through a scaled logit link function which imposes some upper limit, c

Summary for Lay Audience

Wildlife conservation has become a global task in the past few decades. A large number of studies and field experiments have been conducted to assist in wildlife management and conservation. Ecologists are often interested in monitoring the abundance or understanding factors that affect survival of an animal population, and have proposed many statistical models to study these properties.

The Cormack-Jolly-Seber (CJS) model with covariates is often used to estimate the survival of individuals in a population and identify factors that affect survival in order to understand how individuals might behave differently and how the population might be affected by changing conditions. Factors that vary between individuals and over time, like body mass, present a challenge because they can only be observed when an individual is captured. Several extensions of the CJS models have been proposed to deal with the missing-covariate problem and to understand the effects that a factor may have on survival even when it can't be observed at all times for all individuals. However, these models all impose the assumption that the survival probability varies between 0 and 1.

In this thesis, I examine the behaviour of the different models when the survival probability is modeled via a scaled logit link function that restricts the survival probability to be less than some constant, c < 1, that must be estimated from the data. In particular, I explore the estimability of four possible models: the binomial model, trinomial model, full-likeilhood model, and alternative trinomial model. Through a combination of theoretical analysis and simulation I show that effects of an indivdiual time-varying covariate on survival cannot be estimated from the binomial model when survival is scaled to be less than some c

Download

Share

COinS