 |
Valen E. Johnson, Ph.D. 1989, The University of Chicago The University of Texas M. D. Anderson
Cancer Center |
Research Interests:
My research is directed toward the application of novel statistical techniques for the analysis of high-dimensional, highly structured stochastic systems. I am particularly interested in developing techniques for determining whether complex statistical models provide an adequate fit to observed data and how to translate information contained in classical testing procedures into more useful summaries of the posterior probability that a hypothesis is true. I am also interested in developing new models for the analysis of ordered categorical data, particularly for data derived from medical diagnostic tests.
Students performing
a tutorial under my supervision would either apply and develop statistical models
for application to clinically important data sets or would participate in the
development of new Bayesian model assessment, diagnostic, or testing criteria.
Selected Publications:
Johnson VE (2008) Statistical analysis of the National Institutes of Health peer review. Proceedings of the National Academy of Sciences, 105(32):11076-11080.
Hu J, Johnson VE (2008) Bayesian model selection using test statistics. J Royal Statistical Society, Series B [to appear].
Hu J, Joshi A, Johnson VE (2008) Log-linear models for gene association. J Am Statistical Assoc [in press].
Johnson VE, Rossell D (2008) Properties of Bayes factors based on test statistics. Scandinavian J Statistics, 35:354-368.
Program Affiliation:
Program in Biomathematics
and Biostatistics