You are here

The use of bootstrap weights for variance estimation

Danielle Forest
Statistics Canada
This workshop will cover the bootstrap method used to estimate the variance of a parameter. The bootstrap is an inference technique based on successive resampling. The survey bootstrap exploits the existing sample to build synthetic samples, called replicates. These replicates are used to estimate the variance of a parameter. For example, this parameter can be a mean, a ratio or the coefficient of a variable in a regression model. Since the estimator is calculated from a sample randomly selected, it follows that the estimator can vary among the different samples that could be generated from the same population. This variation is expressed by the variance and the variance reflects the reliability of the estimator. Both the estimator and the margin of error are needed to generalize results to the population (inference). This workshop will allow participants to gain the theoretical concepts of the survey bootstrap method and of the variance estimation of a parameter. In addition, the participants will learn to use different software for estimating variance by using bootstrap weights (SAS, STATA and SUDAAN).
Where: 
McGill University
Date: 
8 February, 2013
Training Categories: