I have noticed that normal population parameters can be estimated accurately with a reasonably small sample.
It occurs to me that this is only possible because of the assumption that the random sample is in fact random. The sample is probably not random.
Is there a method of estimating the randomness of an SRS?
