Chp 12 Stats

observational study

indiviudals not assigned, observed only, two types retrospective and prosepective

prospective study

collect data as study goes on

retrospective study

collect data after individuals have made it

experiment

randomly assign individuals to treatments before observation, try to est. cause and effect relationship, different conditions assigned to experimental units

special about obeservational

no control over participants, not random, can only establish possible relationship

special about experiment

some level of control by admin, randomization required, attempts to establish relationship between variables

experimental unit

individuals assigned to treatments, ppl referred to as subjects/participants

factor

explanatory variable (time, medication, location,etc) whose levels are manipulated intentionally

level

specific values chosen for facotr

treatment

the levels or combinations of levels of the explanatory variables

response

the outcome from experiemental units thats measured post treatment

confounding variables

related to explanatory variable and influences the response variable and may create false perception of association

comparison

AT LEAST two treatment groups, one may be control group

randomization

random assignment of treatments to exp. units, helps eliminate unforeseen variation

replication

more than one exp unit in each group, allows for better generalization of factors/levels

control

potenial confounding variables when appropraite, makes sure conditions for all groups similar as possible

blocking

unses explanatory variable to break down participants BEFORE randomization

statiscally significant

difference in effects of treatments not by chance alone

experiments

try to assess effects of treatments

samples

help estimate parameters

control group

group not given treatment or given inactive treatment, own level and compare results to

blinding

keeping treatments mystery, two groups - those who can influence results, those who evaluate results

single blinding

subjects don’t know what receiving, but reserach team does (vise versa)

double blinding

neither subjects or members of research know which treatment is being received

placebo

fake/inactive treatment

placebo effect

those taking placebo think getting treatment report effects of treatment when nothing is happening

randomized blocking design

• subjects split on preexisiting quality

• treatments assigned randomly within each block

• ensures beginning of experimental units similar with at least one variable

• helps seperate natural variablitiy

matched pairs block design

each participant paired with another who is similar in as many ways as possible EXCEPT varibale of choice

confounding

levels of one factor assocaited with factor of another, usually difficult to make conclusion about factor

suvey/observe vs experiments

survey/observational studies only conclude an association, experiments have a causal conclusion