Biostats Flashcards
when 2 diagnostic tests for a dz are compared and one test diagnoses the dz earlier, but there is no effect on disease outcome
lead-time bias
nonrandom sampling or treatment allocation of subjects that study popn is not representative of target popn
selection bias
study popn selected from hospital is less healthy than general popn
Berkson bias
participating subjects differ from nonrespondents in meaningful ways
non-response bias
awareness of disorder alters recall by subjects
recall bias
information is gathered in a systemically distorted manner
measurement bias
when group being studied changes their behavior in order to meet researchers expectation
Hawthorne effect
when an extended period of continuous exposure may be necessary before an outcome may be seen
latent period or length-time bias
if the prevalence of a disease is low, then the OR approaches the RR
rare disease assumption
subjects in different groups are not treated the same
procedure bias
example: patients in treatment group spend more time in highly specialized hospital units
procedure bias
researchers belief in the efficacy of a treatment changes the outcome of that treatment
observer-expectancy bias (Pygmalion effect)
when a factor id related to both exposure and outcome, but not on the causal pathway
confounding bias
early detection is confused with increased survival
lead-time bias
mean > median > mode
positive skew with tail to the R
