ABSTRACT
A significance test, or test of hypothesis, is another. Rather than specify a range of values for a population parameter, a significance test assumes a value for the population parameter and then computes a probability based on a sample given that assumption. The ideas behind a significance test can be illustrated by analogy to a criminal trial in the United States–as seen on TV. Imagine the following simplified scenario: a defendant is charged with a crime and must stand trial. The performer of a significance test seeks to determine whether the null hypothesis is reasonable given the available data. The evidence is replaced by an experiment that produces a test statistic. By specifying a significance level, we indirectly find values of the test statistic that will lead to rejection. This allows to specify a rejection region consisting of all values for the observed test statistic that produce p-values smaller than the significance level.
