STATS 7 Lecture Notes - Lecture 9: Blind Experiment, Repeated Measures Design, Observational Study
Document Summary
Placebo effect: an improvement in health when the patient is under the impression that they are under care (medical, usually) In an observational study you usually can"t take data from the full population: lack of realism: conclusions only apply to individuals like the ones studied, underrepresentation: its hard to apply findings to underrepresented people groups. Slide 12: experimental designs, completely randomized: each individual is studied under only 1 condition, determined randomly. Half of all are given a first, then b; the other half is given b then a. A lot of noise exists between groups (ie tastes, favorites, etc), so having the same person rate both, you can see a trend easier. Issues with the study: it might be more likely that people just choose the first bowl they try, they may be full when trying the second one etc. But because you spread out the groups, it evens out the advantage so both groups have equal chance.
