What do you do if your sample of size $n$ is too small (i.e. <30) ?

What if you care about a statistic other than mean, such as median, quartiles, variance, etc.

BOOTSTRAPPING

A way to estimate a sampling distribution via resampling

• Sample with replacement from your original sample
• Calculate your statistic from this “new” sample
• Repeat

HOWEVER, may not be very reliable if:

• You’re estimating a statistic that heavily depends on a fewvalues, especially if those are extreme values
• Your sample’s distribution has heavy tails (making outliers common)