The need for such procedures arises because the traditional methods based on least squares, such as the T and F tests, are not very efficient for some of the more practical underlying models. The traditional methods are excellent when the model is normal but seem to be quite sensitive to that assumption in the sense that there are much better methods available when the model is not normal. These traditional methods are also not robust; it takes only a few bad observations to disrupt their inferences.
Specific attention has been concentrated on the development of tests with bounded influence and positive breakdown point for subhypotheses in the linear model. Further interest is in structured subhypotheses that might be expressed by an ordering of the parameters. Development of diagnostics for assessing the quality of robust fits in linear models is also very important.
The extension of the idea of the median to higher dimensional data and the definition of sign and rank tests is another area of Dr. Hettmansperger's research. Primary interest is in new methods that are invariant to linear transformations of the data. The development of these methods for more complex experimental designs is another part of the project.
Dr. Hettmansperger was head of the statistics department from 1988 to 1990. He is a Fellow of the American Statistical Association, Institute of Mathematical Statistics, and an elected member of the International Statistical Institute. He was the 1986 recipient of the C. I. Noll Award for Teaching in the Eberly College of Science. He has held visiting appointments at the University of California, Berkeley, Princeton, University of Washington, Latrobe University in Melbourne, Australia, University of Melbourne, The Australian National University, The University of New South Wales in Sydney, Australia, University of Bern, Switzerland, and The University of Oulu, Finland.
Representative Publications:
Robust Nonparametric Statistical Methods, with J.W. McKean, 1998, Arnold/Wiley, London and New York.
Sheather, S.J., McKean, J.W. and Hettmansperger, T.P. Finite Sample Stability Properties of the Least Median of Squares Estimator. Statist. Computations and Sim., 1997, 58, 371-383.
Brown, B. M. and Hettmansperger, T. P. Normal Scores, Normal Plots, and Tests for Normality. J. Am. Statist. Assoc., December, 1996, 91, 1668-1675.
Hettmansperger, T. P. The Ranked Set Sample Sign Test. J. Nonpar. Statist., 1995, 4, 263-270.
Brown, B. M., T. P. Hettmansperger, J. Nyblom, and H. Oja. 1992. On certain bivariate sign tests and medians. Journal of the American Statistical Association, 87:127-135.
Hettmansperger, T. P., and S. J. Sheather. 1992. A cautionary note on the method of least median squares, American Statistician, 46:79-83.