Friday, November 15, 2019

Bootstrap Weights with the CCHS 2015-2016


A researcher is using the Bootstrap weights with the CCHS 2015-2016 and has the following question:

“I am conducting data analysis in STATA using the bootstrap weights, and one of the parameters for the function is bsn, which "specifies that # bootstrap replicate-weight variables were used to generate each bootstrap mean-weight variable specified in the bsrweight() option of svyset. The default is bsn(1)".  Do you know what the bsn number is for the CCHS 2015-2016 file?  I found documentation from CCHS 2013 which stated the bsn number was 500, but I cannot find any similar documentation for CCHS 2015-2016.”


For the 2015-2016 CCHS file there are 1000 bootstrap replicates, this should be for your degrees of freedom (DOF) in STATA.

For the BSN, the CCHS uses the Rao-Wu-Yue bootstrap to estimate the variance. In this variant of the bootstrap, a single bootstrap weight is used to generate each bootstrap weight.

So the default value of 1 should be used. ( BSN(1) )