What is Performance Bias?
Performance bias happens when one group of subjects in an experiment (for example, a control group or an treatment group) gets more attention from investigators than another group. The difference in care levels result in systematic differences between groups, making it difficult or impossible to conclude that a drug or other intervention caused an effect, as opposed to level of care. A similar bias is verification bias, where outcomes are more likely to be found in treatment groups due to investigators knowing which person is in which group.
Performance bias can also refer to the fact that participants can change their responses or behavior if they know which group they are allocated in. For example, if a weight loss study is investigating if a high protein diet works to reduce weight, participants might up their protein intake. This particular type of bias is also called the set of Hawthorne effects.
Performance bias is a major threat to internal validity. Internal validity is a measure of how sound your results are, or how confident you are that you the result of your experiment are due to a single independent variable.
If the investigators know that an experimental group have been given an active drug, they may focus their attention on this group. The participants might receive more frequent exams and more diagnostic tests. This could result in the experimental group having a greater chance of a positive outcome — not because they have been given an active drug, but because they received very focused attention.
Some specific examples of performance bias:
- “Performing tonsillectomy with electrocautery in the group receiving antibiotics while performing cold tonsillectomy in the control group” ~ Bluestone.
- A 1975 trial showed that ascorbic acid had a positive effect on the common cold. This was later found to be due to patients knowledge of the treatment (Karlowski et. al).
Eliminating Performance Bias
This type of bias is more likely to happen if investigators know which group a participant is in. It can be minimized or eliminated by using blinding, which prevents the investigators from knowing who is in the control or treatment groups. If blinding is used, there still may be differences in care levels, but these are likely to be random, not systematic, which should not affect outcomes.
Blinding is not always possible or ethical. For exampled, there is some intense debate about whether any kind of sham surgery is acceptable as a control.
Bluestone, Charles. Pediatric Otolaryngology. PMPH-USA, Jun 1, 2013.
Karlowski, T. R, T. C Chalmers, L. D Frenkel, A. Z Kapikian, T. L Lewis, and J. M Lynch. 1975.“Ascorbic Acid for the Common Cold.” JAMA: The Journal of the American Medical Association 231 (10): 1038–1042.
Sihoven, R. et. al. Arthroscopic Partial Meniscectomy versus Sham Surgery for a Degenerative Meniscal Tear. N Engl J Med 2013; 369:2515-2524December 26, 2013DOI: 10.1056/NEJMoa1305189------------------------------------------------------------------------------
If you prefer an online interactive environment to learn R and statistics, this free R Tutorial by Datacamp is a great way to get started. If you're are somewhat comfortable with R and are interested in going deeper into Statistics, try this Statistics with R track.Comments? Need to post a correction? Please post on our Facebook page.