13 - Research bias Flashcards
Why is it important to understand the effects of bias in research?
Because bias can cause:
- lack of objectivity
- form systematic errors
- create influences: opinions, decisions, conclusions
A form of systematic error in the research process that consistently shifts results in one direction:_______________
Bias in research
T or F: Bias can occur in all phases of research: planning, data collection/intervention, data analysis and publication
T
Some degree of bias is __________ in a published study. However, a study ____________ according to the degree of bias
always present
loses validity
T or F: To understand how bias might influence a study’s results and conclusions is essential for evidence-based practice
T
Name the types of bias in Research planning:
- Design bias
- Selection bias (self-selection, convenient sample, non-response)
A flawed study design is a __________
design bias
Selection bias occurs during the identification of a ______________________
study population
Individuals are more likely to be selected for a study than others =
selection bias
Selection bias is a particular problem in which types of studies:
case control and retrospective cohort study
What could be an effect of participants volunteering to participate in a study?
1) Self-selection or convenient sample
- Individuals select themselves into a group
- Characteristics of participants can affect the results
2) Non-response
- Consider the response rate when conducting a study
How can we avoid self-selection and convenience sampling?
By doing a random selection or random sampling
Each individual is chosen entirely by chance and each member of study population has an equal chance of being included in the sample:
random sampling
How can we avoid non-response ?
- Keep survey short
- Provide incentives for completing the survey
- Explain why this survey is important
What are the types of bias related to data collection/intervention?
- Measure bias
- Interviewer bias
- Recall bias
- Compliance bias
- Attrition bias
Occurs when the measurement tool is not sensitive enough to detect the exposure or outcome of interest:
measure bias
How to avoid measure bias?
Using validated measurement tools that have a high sensitivity and specificity to minimize bias
Systematic difference between how information is solicited, recorded or interpreted:
Interviewer bias
How to avoid interviewer bias?
Ask close-ended questions that are easy to answer vs open-ended questions
Other things you can do to minimize interviewer bias:
- Training interviewers
- Standardize protocols for data collection
- Best if the interviews are blind to the experimental group and control groups
When measuring the outcome or the exposure requires individuals to remember past events: ____________
Recall bias
T or F: Outcome of treatment (good or bad) may change recollections of events
T
Give some examples of diet assessment methods:
- Food frequency questionnaire (FFQ)
- 24-hour recall
- Dietary history
How can we avoid recall bias?
- Giving the participants enough time to reflect before answering
- Using a well-structured questionnaire that has been validated against direct measures
- Verifying the reported-data by using a reference criterion (ex: medical records, data from another family member)
Occurs when some subjects do not adhere to the planned treatment (experimental studies): _____________
Compliance bias
How can we avoid compliance bias?
- Designing protocols that are simple and easy to follow and comply with
- Identifying subjects who are unable or unmotivated to comply (with the intervention) during the enrollment process)
- Contacting subjects frequently to maintain interest and motivation
Occurs when subjects who leave the study (drop-outs) differ significantly from those that remain (common in cohorts and RCT):
Attrition bias
How to avoid attrition bias?
- Using protocols to contact participants by email, post, telephone, etc
- Providing incentives (monetary gifts)
- Keeping the interviews short
What are the 2 types of bias in data analysis and publication:
- Confounding bias
- Publication bias
Occurs when an estimated association between exposure and outcome is biased by a third factor associated with both exposure and outcome: ___________
confounding bias
How can we avoid confounding bias?
- Identified confounders can be controlled with study design (case-control or randomization) and during data analysis (regression analysis)
- Unidentified cofounders can only be controlled with randomization during participants selection and enrollment
T or F: positive results are more likely to be reported and accepted for publication in scientific journals, than negative results
T
Occurs when researchers and research sponsors are unwilling to publish unfavorable results: ______________
Publication bias
It reflects a researcher’s confidence that bias is minimized or eliminated as much as possible so that the results are representative of the outcomes: __________________
Internal validity
Need to consider differences between source population and study population: ________________
external validity
What is external validity:
Degree to which findings are able to be generalized to other groups or populations
