L3 Bias Flashcards
selection bias
choosing participants who may influence the study results
how to avoid selection bias
using random sequence generation and allocation concealment
random sequence generation
researchers randomly assign participants into groups, should be computer generated
allocation concealment
the person randomising the participants to experimental and control groups could not predict what group the person will be allocated to
performance bias
when participants know which group they are assigned to, it can change their responsesor behaviour and affect the outcome of the study
how to control for performance bias
conduct a double blind study
detection bias
when the person or people measuring te outcomes knows which participants received the intervention
how to control for detection bias
conduct a double-blind study
attrition bias
the differences between experimental and control groups because of withdrawals from the study
how to contol for attrition bias
- Explain the reasons for withdrawal and account for all the data
- Authors must do an ‘intention to treat’ analysis - analyse all the data available from the study cohort – those who enrolled at the beginning.
what is reporting bias
- Differences between reported and unreported findings
- Statistically significant differences between intervention groups are more likely to be reported than non-significant differences
- Emphasising positive findings while downplaying negative findings
how to spot reporting bias
when there are multiple research questions outlines and yet the study does not report an answer for all of them
most common type of bias
reporting bias
placebo effect
the participant feeling improved even though they recieved a placebo
nocebo effect
participant feeling disimproved even though they haven’t taken anything (see coke contamination story)
