ap stats - unit 3 vocab Flashcards
Simple Random Sample (SRS)
1) Label 2) Randomize 3) Select
Stratified Random Sample
Split population into groups and choose a SRS from EACH group
Systematic Sample
Randomly choosing a start position, then sample with an equal interval between individuals
Cluster Sample
Split the population based on LOCATION. Use SRS to choose group(s), then sample ALL individuals in those group(s)
Voluntary Response Sample
people who volunteer themselves to respond (due to their own general appeal)
Convenience Sample
choosing individuals who are easiest to reach
undercoverage
when some groups in the population are left out of the process when choosing the sample
nonresponse
when individual chosen for sample can’t be contacted or refuses to participate
wording of questions
the most influential on answers, if worded wrong, could cause more bias
response bias
a systematic pattern of incorrect responses, due to wording of question, LYING, etc…
population
a group of individuals that you want information about
sample
a subset of the population
observational study
observes individuals and measures variable of interest but DOESN’T CHANGE anything to influence responses
experiment
imposes some treatment to measure their responses, doing something to CHANGE the results
explanatory variable (x)
is used to predict values
factors
when there is more than 1 explanatory value
levels
when there is more than 1 factor
response variable
the outcome(s) recorded
treatment
what is being done to the participants
experimental units
are the who/what the treatment is imposed on
confounding variable
the variable not control, but can affect the results
control group
the group with the treatment that has zero level used (like a sugar pill)
provides BASELINE DATA
placebo effect
when a “fake” treatment works on a subject (like a sugar pill)
single blind
when subjects are unaware of what treatment is being assigned (gets rid of placebo effect)
double blind
when BOTH subject and the experimenter don’t know what treatment is being assigned (gets rid of placebo effect)
4 key principles of experiments
(what we can CONCLUDE)
Comparison - need 2 or more treatments
Random assignment - use a chance process
Control - keep all other variables besides treatments constant
Replication - Use enough experimental units to see differences
_______
CRCR
Completely Randomized design
(similar to SRS)
treatments are randomly assigned to a group of experimental units
Randomized block design
when experimental units are grouped by some commonality FIRST, then randomly assigned to a treatment
matched pair design
VERY SIMILAR experimental units paired up at first, then randomly assign treatments (also can be done with the pair being one person, aka most similar you can get)
Causation*
changes in x CAUSES changes in y
Confounding*
changes in x causes changes in y BUT y also is caused by confounding variable z
Common Response*
changes in BOTH x and y are caused by confounding variables b
Statistically significant
a result is unlikely due to chance
