How to Avoid Bias

A valid study is one that is accurate and without bias. Bias may occur in a study in four major ways. Susceptibility bias may arise if groups are not similar at the initial state. The potential for performance bias exists if procedures are not performed comparably. Detection bias may occur if outcomes are not measured comparably. Transfer bias occurs if there is differential loss at followup. The major advantage of a randomized clinical trial is in its attempt to prevent susceptibility bias. Despite the methodologically correct implementation of a study, however, results may not be generalizable due to the particular population, restrictive eligibility criteria, or poor participation. An investigator must balance dual goals: conducting a valid trial and providing generalizable results.