Empirical distributions of parameter estimates in binary logistic regression using bootstrap

Bootstrapping is a famous statistical tool that involves resampling procedure to select sample from a population. In this study, we applied random-x bootstrap in binary logistic regression for published data set namely Umaru Impact data. We conducted bootstrap for the coefficient by using SAS (Stati...

Full description

Bibliographic Details
Main Authors: Fitrianto, Anwar, Ng, Mei Cing
Format: Article
Language:English
Published: Hikari 2014
Online Access:http://psasir.upm.edu.my/id/eprint/37433/
http://psasir.upm.edu.my/id/eprint/37433/1/Empirical%20Distributions%20of%20Parameter%20Estimates.pdf
Description
Summary:Bootstrapping is a famous statistical tool that involves resampling procedure to select sample from a population. In this study, we applied random-x bootstrap in binary logistic regression for published data set namely Umaru Impact data. We conducted bootstrap for the coefficient by using SAS (Statistical Analysis System). We observe the distribution of the estimated coefficients with different sample sizes. After conducting B=10000 bootstrap replications, we found that the distribution of parameters estimates is nearly normal.