bias and confounding Flashcards
internal validity
study representative of reality of study participants
external validity
are the results representative of reality of other populations and at other times
reliability
chances of getting similar results after repeating the test
how to prevent random errors
well planned study
adequate sample size
quality check
measurement tools
random error
occurs randomly cannot be predicted
selection bias
systemic error in the sampling process
information bias
error in the process of collecting and analyzing info
differential if linked to exposure
non differential
how to prevent selection bias
Representative sample
Well written, clear protocols
Criteria for inclusion and exclusion well documented and understood
Minimise attrition
How to prevent information
Bias
exposure and outcome criteria must be clear
adequate staff training
validate data, test tools and conduct a pilot study
subjects and observers should be blinded
how to control confounding during study
randomization
matching - individual matching, cases and controls have the same confounders
restriction
how to control confounding during study
stratification
standardization
multivariable regression