Biases Flashcards
Selection bias
Nonrandom sampling of participants. ex, selected based on adherence to medicines or other criteria that can affect the outcome of the study.
Includes Berkson bias
Health worker effect
Non-response bias
Berkson bias
Study population selected from a hospital is less healthy than the general population
a type of selection bias
Healthy worker effect
The study population is more healthy than the general population
a type of selection bias
Non-response bias
The subjects that participated are different from the non-responders in important ways.
a type of selection bias
Recall bias
Seen in retrospective studies.
Awareness of the disorder affects peoples recall. After learning more about their disease they are more likely to ‘remember’ significant exposures.
Minimize this by decreasing the time from exposures to follow up.
Measurement bias
Information is gathered in a non-standardized manner. Different criteria at different times/places.
Hawthorne effect is part of it.
People behave differently when they know they’re being observed.
Procedure bias.
Subjects in different groups are not treated the same.
ex. people in the treatment group spend more time in specialized hospital units than those in the non-treatment group.
Reduce by using Blinding of patients and researchers and Placebos.
Observer-expectancy bias
Th researchers beliefs about the treatment affect the outcome. If he expects recover he documents it more often.
Reduce by using blinding and placebos.
Confounding bias
When an outside factor is related to both the exposure and the outcome.
ex. coal workers get lung cancer more often but coal workers also smoke more often.
Reduce by repeating studies.
Crossover studies with subjects being their own controls
Matching patients with similar characteristics in the control and treatment/exposure group
Lead-time bias
Early detection is mistaken for longer survival.
