Statistics is about building tools for data analysis.
Statisticians are both tools and tool makers.
If modern Universities had optics departments for
making telescopes and microscopes,
Statistics would fit comfortably in the same division.
Statistics and computer science
are natural allies. Computer science exists because it
makes tools for people to apply
in many different situations. Statistics makes tools
for people to apply to data analysis,
statistical inference, decision making and prediction.
Statistics exists to be applied to data. If it were not
for applications, there would be no point to Statistics
as a discipline. Statistics is a separate discipline and not a
sub-discipline inside other fields because the tools that we
develop are useful across many disciplines.
I have both biostatistical projects and collaborative research projects.
My biostatistical research is developing Bayesian methods for
modeling data. I like to develop statistical models, Bayesian
computing algorithms, and model selection techniques. I particularly
like figuring out the issues involved with analyzing longitudinal data.
I am moving in to semi-parametric Bayesian modeling and multivariate
longitudinal data analysis. In the past, I have developed statistical
diagnostics for residual, influence and sensitivity analysis. I work
on statistical graphics, prior specification and hierarchical models.
Most of my papers combine several of these topics.
Collaborative or applied research is where statisticians apply their expertise to data analysis with the researchers who collected the data.
My applied research involves data and problems from bioinformatics
and community health, mostly involving the HIV virus in some way or other.
In bioinformatics, I have worked on phylogenetic problems, a solution to the
phenotype/genotype problem, detection of selective sweeps and analysis of
the fossil record.
Community health issues that I have applied my statistical expertise to usually involve longitudinal data. One project with Charlotte Neumann and colleagues involves analysis of nutritional, cognitive, anthropometric, classroom, and other data from a school lunch intervention in Kenya. Projects with the Center for Community Health in the Neuropsychiatric Institute involve analysis of data from HIV+ individuals, and individuals often at risk for HIV. I particularly like to work with data from psychometric scales, and self-reported drug use and sexual activity individuals. Most of my biostatistical work involves developing models based on analysis of these data sets.
I especially enjoy teaching my classes in Statistical Graphics,
Longitudinal Data and Applied Bayesian Analysis. I have recently
completed a text book Modeling Longitudinal Data (2005)
published by Springer to be used with my longitudinal data class.
Most of my classes have computing and
writing components. There isn't much point to statistical practice
unless you can calculate what you need to on the computer and communicate
what you have learned. I particularly like working in an academic
environment because I get to work with a variety of subject matter
researchers and Biostatistics graduate students on research projects.
Updated July 2006.