bias, chance and confounding Flashcards
3 types of error
bias
random error (chance)
confounding
what is bias
systematic deviation from the truth
consequences of bias
underestimate or overestimate the parameter you are trying to measure
incorrect estimate of association between disease and exposure
important to identify bias at study design stage as cannot be adjusted for or made allowances at analysis stage
what is selection bias
systematic differences in characteristics between those who take part in a study and those who do not
selection bias in case control studies
cases selected for the study are not representative of all eligible cases
controls are not representative of the population which produced the cases
selection bias in cohort studies
if the comparison groups (exposed and unexposed) and not truly comparable
selection bias in cross sectional studies
participation rarely 100%
individuals who take part may have different exposures
available data on non-responders should be examined
what is information bias
any error in the measurement of exposure or outcome that results in systematic differences in the accuracy of information collected between comparison groups
- reporting bias
- observer bias
reporting bias
when subjects with a specific health outcome report previous exposures with a different degree of accuracy to those without the outcome
decision rule
if p<0.05 reject chance (real effect)
if p>0.05 cannot exclude chance
how to calculate the confidence interval
mean +/- 1.96 * Std error ????
defining confounding
the observed association between two factors is due to the effect of a third factor
- an apparent association may be spurious
- a real association may be obscured
how to deal with confounding
restrict recruitment
- to one level of confounding factor
- could compromise sample size
matching
stratified analysis
direct and indirect standardisation
statistical modelling
