study validity: accuracy Flashcards Preview

DIT topics > study validity: accuracy > Flashcards

Flashcards in study validity: accuracy Deck (13)
Loading flashcards...

systematic errors in study design
flaw in way study is conducted that produces an outcome that differs from underlying truth
false association between risk factor and outcome
most commonly seen in selection of subjects and conduct of study



difference in way study groups are chosen
more patients with end stage disease are referred for the trial than early stage because that's all who is available right now

selection bias


only use hospital patients as study participants - scews association between risk factor and disease since the only participants were hospital patients instead of people in the community too

berkson's bias


difference in recall between 2 study groups
difference in accuracy and detail of history of 2 groups (parents of autistic children vs normal children)

recall bias


study participants are not true reflection of target population (sample group age is different from target population age)

sampling bias


information gathered too late to draw conclusions about the disease
survey needs to be sent out at time of diagnosis to get better sampling of patients with early stage/late/aggressive/indolent disease

late-look bias


study groups in same study not treated the same (controls need to get same attention/education as experimental group)

procedure bias


early detection of disease with screening tool - doesn't change natural history of disease even though cancer survival after diagnosis is increased

lead-time bias


researchers expectations of outcome effect study results
investigator is so confident in efficacy of a treatment that they influence the study subject or test results to provide conclusions that they already expect - more likely to document positive outcomes

pygmalion effect


change in behavior in a study group that can result from knowing they are being observed (wash hands more)

hawthorne effect


systematic error in study design
association between exposure (asbestos miners) and outcome (lung cancer) is distorted (magnified or reduced) by a third variable (smoking) which is associated with the exposure (asbestos minors like to smoke more than other populations) and outcome (smoking causes lung cancer)- all of increased risk for lung cancer may not be due to asbestos exposure, but consider smoking too!



study results occurred due to pure chance
can't be avoided

random error


study design that reduces bias and confounding

randomized, controlled double-blinded trials

Decks in DIT topics Class (115):