On variance estimation and a goodness-of-fit test using the bootstrap method

Abstract: This thesis deals with the study of variance estimation using the bootstrap method, including the problem of choosing between nonparametric and parametric bootstrap methods. Paper I compares the two approaches, determines which method is preferable and analyses the accuracy of the approximations. The underlying concept of parametric bootstrap is based on the assumption of correct choice of parametric distribution. Paper II therefore considers goodness-of-fit tests and presents a new test based on the bootstrap method.