Thomas Severini Professor of Statistics
My research focuses on two areas of statistical theory and methodology: likelihood-based statistical methods and inference in models with infinite-dimensional parameters. Likelihood-based statistical methods include maximum likelihood estimation and tests and confidence regions based on the likelihood ratio statistic. My work in this area is concerned with higher-order asymptotic approximations to the distributions of likelihood-based statistics and with the construction and properties of marginal and conditional likelihood functions. Models with infinite-dimensional parameters include models containing an unknown regression function and models containing an unknown distribution or density function. My recent work in this area involves inference in nonparametric linear models with endogenous regressors and the relationship between models with an unknown regression function as a parameter and random effects models.
- Modified Estimating Equations, Biometrika, 89 (2002), 333-343.
- A Simplified Approach to Computing Efficiency Bounds in Semiparametric Models (with G. Tripathi) Journal of Econometrics 102 (2001), 23-66.
- Likelihood Methods in Statistics, published by Oxford University Press (2000).
- The Likelihood Ratio Approximation to the Conditional Distribution of the maximum Likelihood Estimator in the Discrete Case Biometrika 87 (2000), 939-945.