Competitive Bias Seeking in Reporting Statistical Estimation

George Mason University
CSI/Statistics Colloquium Series
Seminar Announcement

Competitive Bias Seeking in Reporting Statistical Estimation

David Levy

George Mason University

ABSTRACT Bayesian theory considers the utility-maximizing behavior of an isolated statistician. When the maximizing statistician is introduced into a social context, the game of statistical policy is created and thus motivates selective under-reporting of results. The literature has considered what a naive jury can conclude from biased estimates from each side's expert witness and concluded that the biases can be offset by a "split the difference reul."

While the biases may indeed be offset by the competitive court process, the statistical equivalent of the prisoner's dilemma is created where competitive "bias seeking" increases the variance of the estimation. Simultaneous equation estimation provides an important context for such bias-seeking since econometric conventions allows selective reporting.

Friday, December 10, 1999
George W. Johnson Center, Assembly Room B
Seminar at 10:45 a.m.
Refreshments at 10:30 a.m.
For the 1999 Fall Seminar Schedule, go to
http:www.science.gmu.edu/statseminars