Bias in research
What is bias in research
A form of systematic error in the research process: Planning, Data collection and intervention, Data analysis and publication, that consistently shifts results in one direction.
Self-selection and convenient sample
Characteristics of those who volunteer (or are referred) to participate in a study are often different from the non-volunteers (not referred).
How to avoid confounding bias
Control identified con-founders w/ study design and during data analysis. Unidentified con-founders can only be controlled with randomization during participants selection and enrollment.
How to avoid compliance bias
Design protocols that are simple and easy to follow and comply w/; identifying subjects who are unable or unmotivated to comply during the enrollment process; contact subjects frequently to maintain interest and motivation.
Design bias and how to avoid it
Flawed study design/ Select an adequate study design for the research question. such as Prevalence for cross sectional study and Incidence and Prognosis for Cohort.
Confounding Bias
Occurs when an estimated association between exposure and outcome is biased by a third factor associated with both exposure and outcome.
Recall bias
Occurs when measuring the outcome or the exposure requires individuals to remember past events.
Publication bias
Occurs when researchers and research sponsors are unwilling to publish unfavorable results.
Compliance bias
Occurs when some subjects do not adhere to the planned treatment.
Attrition bias
Occurs when subjects who leave the study (dropouts) differ significantly from those that remain.
Measure Bias
Occurs when the measurement tool is not sensitive enough to detect the exposure or outcome of interest.
Interviewer bias
Occurs when there is a systematic difference between how information is asked, recorded, or interpreted.
How to avoid non-response
Provide incentives for completing survey, Explain why survey is imp, and Keep survey short.
How to avoid self-selection and Convenient sample
Randomly select the sample. Make sure each member of the study population has an equal chance of being included in the sample.
Selection Bias
Situation where some individuals are more likely to be selected for the study than others.
How to avoid interviewer bias
Standardizing protocols for data collection, training interviewers, binding the interviewers to the study hypothesis.
Non-response
Those who did not respond/participate often have different characteristics from those who did respond/participate.
What is essential for evidence-based practice
Understand how bias might influence a study's results and conclusions.
How to avoid attrition bias
Use protocols to contact participants by email, post, SNS etc, provide incentives, and keep the interviews short.
How to avoid measure bias
Use validated measurement tools that have high sensitivity and specificity to minimize bias.
How to avoid recall bias
Using a well-structured questionnaire that has been validated against direct measures. Giving the participants enough time to reflect before answering.