In
statistical or
mathematical terms, bootstrapping refers to a
method of determining the behaviour of a
model or a
distribution through
simulation. It is one of the
Monte Carlo methods of simulation, along with
permutation and
jackknifing.
Using a real data series, bootstrapping allows a researcher to create a large number of hypothetical cases by resampling the data set with replacement. The fundamental procedure is as follows:
- Identify the data series of interest. This data series will have n observations.
- Calculate the parameter(s) of interest on the original data.
- Select n random observations from the original data series, with replacement. This means that after selecting a case from the data, this case returns to the pool from which observations are randomly chosen. Thus, a single object may be selected more than once in any single simulation.
- Calculate the parameter(s) of interest on this simulated data set.
- Repeat steps 3-4 as many times as desired (normally many thousands of times).
This method of simulation has become increasingly
appreciated and utilized in modern science. The reason for this is that it allows for the relaxation of many of the
assumptions inherent in most theoretical distributions (such as the
normal distribution,
Poisson,
Chi-squared, uniform etc.).