ABSTRACT
Whenever a statistical test is used more than once, the likelihood of achieving a statistically significant result is increased, even though the Null Hypothesis of no effect is correct. That is, there is an increased danger of making a Type I error. It is possible to adjust the α-level that a given test would have to achieve before statistical significance was considered to have been reached, to allow for the number of times the same test was being conducted. A general method is described as the Bonferroni adjustment.
