The bootstrap is a resampling method to estimate the sampling distribution of a statistic.
This is what the tag description says on this website. I am a layman, can someone please help me understand what is bootstrapping with an example?
Asked
Active
Viewed 440 times
1
-
1The Wikipedia entry is not bad (although perhaps more detailed than you wanted): http://en.wikipedia.org/wiki/Bootstrapping_(statistics) – Jeremy Miles Nov 22 '14 at 18:20
-
Thanks. They show a coin toss example and it seems that the second sample of X1,x2....x10 is basically almost the same as the first sample in a different order. I was expecting the second sample to be from X11,x12.....x20. Is not the second sample supposed to have different data points from the first? Or are they allowed to overlap? And to what degree? – Victor Nov 22 '14 at 19:31
-
But they are resampling from the same 10 observations in that example. Overlap is a consequence of sampling with replacement. In other words, imagine a bag of white, red and black rocks (say, 50:100:150), you take one rock at a time, record it's color and throw it back in the bag. Do it 100 times, what's the ratio you get? That's your bootstrapped with resampling estimate. – katya Nov 23 '14 at 21:05