Until now, we have using methods that assume that variable has a distribution with some characteristics: a) the distribution is Normal, in quantitative data, and b) in binary data, the distribution is binomial, with a probability, p.

Means, standard deviations and probabilities, are calling parameters and the methods to make inferences about these parameters, are calling parametric methods.

There are other distributions; with them, we do not assume that they are Normal or binomials; as when the sample size is small, quantitative data do not have Normal distribution or when categorical data have more than two values.