102 Statistics Building
Ft. Collins, CO 80523
(970) 491-6786 - phone
(970) 491-7895 - fax
jbreidt@stat.colostate.edu

Courses

Reports & Papers

Short Vita

Vita (PDF format)

Lectures

Biography

Current Research

Other Interests

Breidt Home

Statistics Home

My research interests include time series and survey sampling. In survey sampling, my interests include regression weighting, variance estimation, and sampling design, often with a view toward application in environmental resource inventories over large landscapes. My professional practice has often involved consultation on the design and analysis of large-scale government surveys. I like to think that this experience has shaped my methodological and theoretical research in the field of surveys. I try to develop methods which have considerable efficiency when everything is "nice", but which do not break down when things are not so nice. Computational tractability in the demanding operational environment of a real survey is something of real concern to me. Lately, I have been working with my colleague Jean Opsomer on nonparametric regression estimators for surveys. I think this has great potential as a toolkit for incorporating cheap auxiliary information (such as remotely-sensed data) into the estimation of population parameters. This work is supported by the National Science Foundation (theory), the US Forest Service (methodology for forest inventory and monitoring), and the US EPA (aquatic resources monitoring and assessment). The EPA funding is through a program called STARMAP (Space-Time Aquatic Resources Modeling and Analysis Program) and through a parallel program at Oregon State. Much of this work fits into the interdisciplinary research supported by PRIMES (PRogram for Interdisciplinary Mathematics, Statistics, and Ecology), an NSF-funded IGERT that offers generous support for graduate students interested in quantitative ecology. I was one of the principal writers of this proposal and I serve on the PRIMES Council which directs the program.

In time series, specific areas of activity for me have included non-Gaussian linear processes, time series methods for repeated sample surveys, stochastic volatility models, and nonlinear filtering. Lately, I have returned to my roots in non-Gaussian linear processes (which I explored in my dissertation) by working with Richard Davis on all-pass models and their applications. These are autoregressive moving-average models with a special structure, which can arise in the identification of non-causal or non-invertible systems. The all-pass methodology yields a combined approach to the identification and estimation of such systems, and potentially offers great efficiency advantages over the moment estimators in current use.