Overview of Honours, MSc, and PhD projects

Students with strong backgrounds in statistics (or related subjects) wishing to pursue a postgraduate research degree in statistics may wish to consider the projects listed below. If so, funding for MSc or PhD students can be applied for through the Otago scholarships office for both domestic and international students. Dr. Dillingham is also available to co-supervise other students at Otago in areas related to his work (e.g. conservation biology, ecology, chemistry).

Experimental design and model-averaging

Model-averaging is one approach to multi-model inference, and has been advocated as a standard method for the analysis of factorial experiments (Fletcher and Dillingham 2011). This project would expand this work to other classical designs and study design implications when using model-averaging. Much of our motivation comes from our research in globabl ocean change studies, described next.

Experimental design for global ocean change research

Closely linked to the project above, this project would pursue development of experimental designs tailored to the needs of ocean change researchers. With a need to study the effects of multiple stressors while constrained by small sample sizes, this project would be ideal for a student interested in the interface of statistics and science. See our poster on model-averaging in split-plot designs for further details.

A related project focusses on analysis tools for mesocosm experiments in conjunction with Cliff Law (NIWA) and Linn Hoffmann (Otago Botany Department). Particular characteristics of these studies are multivariate responses with small sample sizes monitored over time.

Measuring overdispersion in Bayesian models

We wish to build a measure of overdispersion in Bayesian models (similar to φ in GLMs) that can be used to provide simple post-hoc adjustments to credible intervals. This project will use simulation to describe properties of several proposed methods.

Estimating HIV incidence in New Zealand (with Dr Jiaxu Zeng)

In order to estimate incidence of HIV in New Zealand, a Bayesian model will be implemented (based on a paper by Sweeping et al. 2005; Statistics in Medicine). This project will extend their model and apply it to the New Zealand context.

Integrating life history theory, matrix models, and abundance time series

The rate of population growth is one of the most important demographic parameters. This research project will explore ways to make use of multiple data sources in order to better estimate growth. New methods will be applied to populations of marine megafauna such as seabirds, sharks, or marine mammals.

Contact Dr Dillingham for details.