Say we have a small number (N=10) of measurements of a variable X from an unknown distribution (possibly skewed or bimodal).
And I want to calculate the probability that another measure belongs to the same distribution, or the p-value.
How can I do it?
Bootstrap is commonly used to calculate the distribution of an estimator, for example the mean.
But in my problem I don't need to study the mean (of samples of a given size) but single values, my sample has size N=1.
We can generate new values in the region of the values I already have in my sample with oversampling methods. But, how can I properly generate new data beyond the values in my sample, in the tails?
