Lecture 29: Some Non-Parametric Tests

To this point, all of our exact tests have been based on knowing fairly specific information about the null hypothesis and alternative hypotheses. (The exception is the Z-test.)

Generally speaking we assume that we know the form of the density or mass function except for a parameter, and the purpose of our test is to decide whether or not a particular parameter value is ``reasonable''.

Now we want to look at something more general, and is some sense, a lot more believable. We know absolutely nothing about the functional form of the population distribution. What kinds of information can we test for?