Bias and variability in clinical research.

Two of the practical problems facing investigators attempting to compare alternative orthodontic treatments are bias and variability. Since the subjects receiving treatments are usually different people, observed differences can be due to differences in people rather than differences in the treatments. If the groups being evaluated differ in any systematic way, they are said to be biased. Treatment comparisons between biased groups are difficult to interpret since bias can exaggerate, nullify or reverse true differences. The random assignment of patients to alternative treatments breaks any systematic connection between treatment and any variable that might favor one treatment over another; thus, making it safer to attribute differences in outcome to the treatments given rather than the people studied.