* An example of creating and using bootstrap weights. * Reference: Valliant, Richard and Jill A. Dever. Survey Weights: * A Step-by-Step Guide to Calculation. 1st edition. College Station, * TX: Stata Press, 2018. pp.97-98. * https://www.stata-press.com/books/survey-weights/ * Get the bsweights package for creating bootstrap weights. findit bsweights * In the Viewer window that pops up, click on link for: * st0187 from http://www.stata-journal.com/software/sj10-2 * In the next Viewer window that pops up, click on: * (click here to install) * Once the bsweights package is installed, get the following data file * to try out creating bootstrap weights: use http://www.stata-press.com/data/svywt/smhosam, clear * Compare descriptive statistics for the number of hospital beds * using 1) no weights, 2) the probability weight only and 3) the * the probability weight and 500 bootsrap weights (to be created). summarize BEDS mean BEDS * Set up the complex samples scheme with the (one) probability weight. svyset ID [pweight=wt], strata(hosptype) svy: mean BEDS * Create the bootstrap weights. An example is given below. * Documentation and command options are available at: * https://www.stata-journal.com/article.html?article=st0187 * and click on: * View this article (PDF) * This article includes an explantion of how bootstrap weights are created. * Please note: As mentioned above, the example given here is from * Valliant and Dever, Survey Weights, 2018, pp.97-98. A seed value is set * in order that the bootstrap weights can be replicated exactly. bsweights bwt_, n(raowu) reps(500) double seed(339487731) * Compare a few of the boostrap weights to the probability weight. summarize wt bwt_1 bwt_2 bwt_499 bwt_500 total wt bwt_1 bwt_2 bwt_499 bwt_500 corr wt bwt_1 bwt_2 bwt_499 bwt_500 * Set up the complex samples scheme with the (one) probability weight * and the 500 bootstrap weights just created. svyset ID [pweight=wt], bsrweight(bwt_*) vce(bootstrap) mse svy: mean BEDS