Health Economics

  • (written as Bootstrapping)
    A non-parametric method of estimating the distribution of an estimator or test statistic by 'resampling' the data. The term comes from the old idea that you might be able to lift yourself off the ground by pulling on the straps on the backs of your boots. Suppose you have a sample of 20. You 'bootstrap', or recreate, the population from which the sample came by duplicating the sample many times over in a computer simulation of the population. Thus, you draw a sample (say of 1) from your sample of 20, record its value, replace it amongst the 20 and draw again. This process is repeated many times. Bootstrapping is particularly useful when data are skewed and sample sizes are modest. It is frequently used in estimating probability distributions of cost-effectiveness ratios, their confidence intervals and variances.