Lecture 7 - Bias Flashcards
define selection bias
sys error in selecting participants; do participants rep pop? are participants similar in all aspects except exposure/disease?
define information bias
data for outcome and exposure obtained differently; incomplete or inaccurate data collection
What are diff terms for information bias
observation or measurement bias
define confounding bias
results accounted for by presence of factor associating with exposure and outcome but not in causal pathway
Selection bias affects which two observational studdies
CC and cohort
What are the 4 types of selection bias
volunteer, non-response, exclusion, healthy worker bias
define exclusion bias
exclusion criter/participant loss after randomization results in a group of people that dont rep gen pop
Exclusion bias can also occur if
diff eligibility criteria for cases and controls
what are the 4 contributing factors of healthy worker bias
healthy hire, healthy worker survivor effect, time since hire, beneficial effect of work
what are the two phases of time since hire?
dynamic and stable phase
what are the 6 types of info bias
non differential bias, differential bias, recall bias, interview bias, LTFU bias, reporting bias
non differential and differential bias are classified as what type of info bias
misclassification bias
define non differential bias
measurement of disease is not diff btw exposed and unexposed and vice versa
define differential bias
measurement of disease is diff btw exposed and unexposed and vice versa
in which misclassification bias does the rate of misclassification diff btw diff study groups
differential bias
how do you combat recall bias
use controls with a disease with similar potential for search as cases, hospital records, biomarkers, blinding
define LTFU bias
systematic difference in following and obtaining outcome info; problem with LTFU relates to outcome
What are two types of reporting bias
social desirability bias and wish bias
define wish bias
deny exposure to minimize self blame
what is another term for LTFU
attrition bias
RCTs have which bias
selection, performance, attrition, detection bias
CC have which bias
recall and selection
Cohort has which bias
selection bias due to diff rates of LTFU in exposed and unexposed
Define random error
reduce precision of estimates; non-differential missclassification
define systematic error
reduce validity/accuracy of estimates; differential miscclassification
in systematic error
you always underestimate a parameter, therefore a predictable distance away from truth
in random error
you on average get right answer but you sometimes over and underestimate
define precision
replicability of value with multiple measurements
define accuracy
getting results right
