In the context of analyzing occupational exposure data, my work is on the development of statistical methodologies that are better suited and more accurate for exposure monitoring in a wide variety of workplace environments. This is critical for setting exposure limits and for assessing occupational risk. The relevant data are usually lognormally distributed, and mixed and random effects models are very often appropriate. The problems of interest here deal with the development of tests and confidence regions concerning one or more lognormal means, the computation of tolerance regions and calibration procedures, and the development of techniques to deal with data below the detection limits. This work has been funded through a grant from National Institutes of Health.
I am interested in all aspects of statistical inference concerning linear mixed and random effects models. My current research interests in this area include the development of tests and confidence intervals concerning `non-standard' parametric functions involving fixed effects and variance components. The problems came up in the analysis of Army test data at the Army Research Laboratory, Aberdeen Proving Ground, Maryland. My ongoing work on the development of tolerance regions, univariate as well as multivariate, extend to mixed and random effects models as well.
Recently I started
investigating the application of higher order asymptotic theory for computing
tolerance intervals in mixed and random effects models. Higher order asymptotics appears to be a
option for computing tolerance factors in very general mixed and random effects models, even if the
unbalanced. Currently I am also investigating higher order asymptotic procedures for biomarker