Chance is vital in statistical design
random sampling and randomized comparative experiments are extremely important and effective statistical practices
experiment
treatment imposed on individuals in order to observe responses
control
overall effort to minimize variability in the way experimental units are obtained and treated
experimental units
individuals on which the experiment is done: people: subjects
treatment
a specific experimental condition applied to experimental units
factor
explanatory variable (treatment)
placebo
a dummy treatment that can have no physical effect
control group
receive dummy treatment (helps experimenter control effects of lurking and confounding variables;
completely randomized experimental design
all experimental units are allocated randomly among the treatments
statistically significant observation
an observed result too unusual to be an outcome determined by pure chance
3 principles of experimental design
1) control: needed to counter effects of lurking/confounding variable; simplest form is comparison (2 or more treatments); types include control group, placebo, blinding; 2) Replication: do experiment on many experimental units to reduce chance variation
types of experimental design
completely randomized design; block design; matched pair design; refer to diagrams in notes
completely randomized design
3 elements: 1) control: administration of different treatments and comparison of the outcomes; 2) replication; 3) randomization: random allocation of subjects to treatment and control groups
block design
group of subjects known before the experiment to be similar in some way that affects response to treatment (age, gender, race, health); perform completely randomized experiment within each block; blocking is form of control; purpose is to control effects
block design mantra
control what you can; block what you can't control; randomize the rest
matched pair design
special case of randomized block design; used when experiment has TWO treatments and subjects can be matched in pairs that are similar, based on some blocking variable; within each pair, subjects are randomly assigned to different treatments
variation of matched pair design
one subject/both treatments on same subject; randomly choose which treatment to do first; measure the difference between treatments; added level of control because the differed between treatments is measured (improvement)
placebo effect
phenomenon where patients get better because they expect the treatment to work even though they have been given a placebo (fake treatment)
control: blinding
subject doesn't know what group they are in; double blind experiments; neither the subject nor the reacher knows what group they are in (controls placebo effect)
ap exam tips
need to be clear on the distinction between the purposes for "blocking" and randomizing; if you are asked to describe an experiment involving blocking, be sure to randomize treatments within the block