Resampling (statistics)

Resampling means taking a small toy set of candies out of a big jar with lots of candies. You pick one candy after another and write down its color or flavor. Then, you put it back into the jar and shake it. This is called one "trial" of resampling. You repeat this several times and write down the colors or flavors each time. This is called "repeating the resampling". Eventually, you have a list of colors or flavors that you can use to learn about the big jar of candies.

This is what statisticians do with data: they take a sample of data out of a big set of data, and do something to it to learn about this big set. For example, to decide whether a new medication works, they can give it to a small group of people (a "sample") and see how they do. Then, they use resampling to see if the new medication would work for other people who were not in the sample.

Overall, resampling is a way of testing ideas about how a big set of data might look like based on a smaller (sampled) set of data. Just like with candies, we can learn a lot about the entire jar of candies by sampling only few of them.