Bias Flashcards
What is Bias?
Any error that is a result of methods used in the study design, from methods that effect study participation to methods that effect collection of information about the exposures and outcomes.
A systemic error that is non-random and deviates from the truth
How can we avoid Bias?
Appropriate study design
What does not effect/avoid Bias?
sample size
cannot be adjusted for in stats
What is Selection Bias?
Who is included in the study sample and how representative they are of the wider population or the group they are being compared to.
What can reduce selection bias?
Random selection
Random allocation
Clearly define groups being compared
Example of Bias in an occupation cohort study?
The Healthy Worker Effect:
Workers appear healthier than non-workers because the non-worker group will include people who are too sick to work.
Reduce by random allocation of workers/non-workers to both groups.
What can lead to selection Bias?
Inclusion/exclusion criteria - pre-screening may lead researcher to choose people who are more likely to respond to treatment etc.
Methods of counting/enumerating - e.g is researchers are given a list of civilians in a village drawn up by elders it may not include recent migrants and so will not be wholly representative of the study pop
Volunteer Bias - people who self-select for a study may have greater health awareness than general population.
What is information bias?
Who can it be introduced by?
AKA asceratainment Bias
when the outcome or exposure is classified inaccurately OR any deviation from the truth when collecting information on the exposure or outcome
Eg. recall etc.
Can be intro’d by researcher, study participants, or measurement tools
What is observer Bias?
Researcher
Caused by researcher missclassifying information on groups.
Eg. rounding up for one group and down for another
How can we avoid observer Bias?
Blinding
- Hiding the exposure status from the person taking the measurements
OR
all measurements taken by the same person
OR
all measurements taken by random investigators.
What is responder Bias?
Differences in information provided by the participant.
E.g No of sexual partners and No of STI’s
What is it called if someone under or over reports?
Reporting Bias
Example, under reporting number of partners due to stigma
What is recall Bias? (part of responder bias)
Person deviating from the truth when recalling facts.
E.g. someone with lung cancer from cigarette smoking vs someone with lung cancer from secondary smoke exposure - the person with SSE may have better recall if they were aware of the connection between the 2 things.
What is response Bias?
If someone chooses not to respond to certain questions.
Eg. people who are less sexually active may not respond to certain q’s due to embarrassment.
How can we reduce responder Bias? or information Bias?
Blinding - e.g. using a placebo in trials or concealing the hypothesis from participants
Asking a question in different ways to ensure consistency of answers
Asking about a number of exposures, vs focusing on one.
Shorten the recall period.
Piloting questions before study
