Bias Reduction of Estimates By Bootstrap Method
Date of Graduation
Master of Science in Mathematics
In general it is desirable to have unbiased estimators for parameters of a probability distribution function. However, there are several estimators which are not unbiased. In this thesis, we show by direct computation that the bias of the bootstrap estimate of μ⁴ can be reduced, where μ is the mean of the population. We consider both nonparametric and parametric cases. In the parametric case, the bias of the bootstrap estimate is computed for normal, exponential and Poisson distributions.
© Linda J. Schuchman
Schuchman, Linda J., "Bias Reduction of Estimates By Bootstrap Method" (1995). MSU Graduate Theses. 1021.