Martin, Michael A2017-11-292017-11-291989b1708855http://hdl.handle.net/1885/136566In this thesis, we discuss the use of bootstrap methods for constructing confidence intervals in a wide variety of circumstances. In particular, we provide asymptotic theory for bootstrap confidence intervals based on Edgeworth expansions in several problems, and develop a general bootstrap resampling scheme applicable to a large range of statistical problem s. The thesis is divided into five chapters. The first gives a brief introduction to bootstrap methods and Edgeworth expansion theory to provide a background and motivation for our later work.183 penSampling (Statistics)Confidence intervalsMathematical statisticsNonparametric statisticsEstimation theory198910.25911/5d70efade2b912017-11-22