Bruce G. Lindsay
Distinguished Professor of Statistics
Ph.D.: University of Washington, 1978
Dr. Bruce Lindsay's Home
Page
Dr. Lindsay's research is in the following four areas.
-
Nuisance parameters: When building a statistical model for data collected
on a sample of individuals, it is typically necessary to include many parameters
to make sure the model is rich enough to approximate the true state of
nature. However, the standard methods of statistical analyses have major
deficiencies when viewed in this framework. Dr. Lindsay has developed a
theory for measuring the deficiency and methods free from the deficiency.
-
Mixture models: If the heights of college students in a classroom are sampled,
then the distribution of the heights will be nonnormal, and possibly bimodal,
because the students are a mixture of males and females. Dr. Lindsay has
developed a better set of tools for the analysis of data from such a population.
-
Computer algorithms: There has been an explosion of new ideas and methods
in statistics because the computer has made it feasible to be much more
sophisticated and realistic in creating models. Dr. Lindsay's work includes
designing computer programs, guaranteed to converge to the right answer,
and on methods for constructing good initial estimates at which to start
the program.
-
Minimum distance and robustness: A long-standing concern in statistical
methods has been the sensitivity of the answers to one or more incorrect
data points; a robust method is one lacking this deficiency. Dr. Lindsay
has been exploring a set of methods called minimum distance and has found
a key to understanding their robustness properties.
Dr. Lindsay was a Woodrow Wilson Fellow (1969), a NATO Postdoctoral Fellow
at Imperial College (1977), and received the Humboldt Senior Scientist
Award (1991) in Berlin. He is a Fellow of the Institute of Mathematical
Statistics.
Representative Publications:
Lindsay, B. G., and D. Boehning, and P. Schlattman. 1992. Computer
Assisted Analysis of Mixtures (C.A. Man): Statistical Algorithms. Biometrics,
48:283-304.
Lindsay, B. G., and K. Roeder. 1992. Residual diagnostics in the
mixture model. Journal of the American Statistical Association, 87:785-795.
Lindsay, B. G., J. Lynch, and H.-L. Hsi. 1992. On mixtures of
hazards: Non-parametric maximum likelihood in a competing risk model. Non
Parametric Statistics 2:89-103.