HMB342H1 Study Guide - Final Guide: Medline, Odds Ratio, Publication Bias
Document Summary
Reliable but not valid: clustered but inaccurate. Not reliable but valid: scattered but accurate. Not reliable and not valid: scattered & inaccurate. Content validity: includes all dimensions of construct and nothing more (charac. of pain) Construct validity: related in a coherent way to other measures that are believed to be part of the same phenomena (constructs of depression) Criterion validity: measures predict a directly- observable phenomena (pain asso. w/ diff. injuries) Do not have control over what exposures were measured nor how: what is the approx. Rr of hpv infection associated with previous hpv immunization: what is the approx. Population incidence = (70% x 220) + (30% x 110) = % of deaths attributable to physical inactivity = Parallel testing: only 1 +ve test is needed; sen. Serial testing: all tests needed to be +ve, sen. (serial: post-test odds = pre-test odds for the next) Sen. -> useful to rule out" disease (-ve test results usually true ve")
