Article not rated

Vol. 16 - Num. 64

Critical appraisal in small doses

The problem of multiple comparisons

Manuel Molina Ariasa

aServicio de Gastroenterología Pediátrica. Hospital Universitario La Paz. Madrid. España.

Correspondence: M Molina. E-mail: mma1961@gmail.com

Reference of this article: Molina Arias M. The problem of multiple comparisons. Rev Pediatr Aten Primaria. 2014;16:367-70.

Published in Internet: 05-12-2014

Abstract

Clinical trials are designed to respond to a well-defined clinical question. However, in many cases, especially if the results are not significant, studies after the test design are established to try obtaining some positive results. Comparisons between subgroups based on nonrandomized characteristics to find a statistically significant result are well established. The problem of multiple comparisons is the increased likelihood of false positives and committing type I errors. For this reason, all the results obtained from the comparison of subgroups formed after randomization should be critically analyzed.

Keywords

Bonferroni Correction Multiple comparisions Post hoc analysis Type I error

 

Comments

This article has no comments yet.