Approximate the Distribution of Statistics
Bootstrapping is a re-sampling method for approximating the sampling distribution of a statistic. The data is re-sampled with replacement and the statistic is computed. This process is repeated to produce a distribution of values for the statistic.
Bootstrapping is useful when estimating properties of a statistic (mean, standard error, and so on) and performing inference, in the following situations:
• The theoretical distribution of the statistic is complicated or unknown.
• The sample size is too small for regular statistical inference.
• Inference using parametric methods ...