Bootstrap Methods in a Class of Non-Linear Regression Models

In this paper, the performances of the bootstrap standard errors (BSE) of the Weighted MM (WMM) estimates were compared with the Monte Carlo (MCSE) and Asymptotic (ASE) standard errors. The properties of the Percentile (PB), Bias-Corrected Persentile (BCP), Bias and Accelerated (BC), Studentized Pe...

Full description

Bibliographic Details
Main Author: Midi, Habshah
Format: Article
Language:English
English
Published: Universiti Putra Malaysia Press 2000
Online Access:http://psasir.upm.edu.my/id/eprint/3511/
http://psasir.upm.edu.my/id/eprint/3511/1/Bootstrap_Methods_in_a_Class_of_Non-Linear.pdf
Description
Summary:In this paper, the performances of the bootstrap standard errors (BSE) of the Weighted MM (WMM) estimates were compared with the Monte Carlo (MCSE) and Asymptotic (ASE) standard errors. The properties of the Percentile (PB), Bias-Corrected Persentile (BCP), Bias and Accelerated (BC), Studentized Percentile (SPB) and the Symmetric (SB) bootstrap confidenceaintervals of the WMM estimates were examined and compared. The results of the study indicate that the BSE is reasonably close to the ASE and MCSE for up to 20% outliers. The BCa has attractive properties in terms of better coverage probability, equitailness and average interval length compared to the other methods.