Unit 3 AP Stats Review

census

census

a procedure by which EVERY MEMBER of a population is selected for study

representative sample

a sample w/ the essential characteristics of the population studied without any bias

sampling frame

list of all the members of population from which we can draw a sample

probability sample

a sample in which every individual in the population has a known probability of being sampled

simple random sample

a sample in which ALL samples of size x have an equal probability of being drawn

systematic sample

a sample in which the starting point is chosen randomly, but afterwards the sample is drawn from some predetermined system/pattern (e.g. every 12th person)

stratified sample

sampling frame is divided into homogeneous groups, then random sample is drawn from each homogeneous group

cluster sample

sampling frame is divided into groups which are all the same/intended to be representative of the population; then, clusters are sampled randomly

self-selected/voluntary response sample

a sample in which people choose whether or not to participate in the survey/respond (TENDS TO BE BIASED, VOLUNTARY RESPONSE BIAS)

convenience sample

sample drawn however is most convenient (TENDS TO BE BIASED)

quota sampling

individuals are selected intentionally to try to reflect different quotas of groups in the population, e.g. selecting x amount of teens to include because teens make up y proportion of the population (TENDS TO BE BIASED)

bias

the tendency of a sample statistic to systematically over- or under-estimate a population parameter (e.g. our sampling method will regularly & systematically produce samples that misrepresent our data in the same way)

undercoverage

one part of the population being sampled is excluded/underrepresented

voluntary response bias/nonresponse bias

certain individuals (those who feel certain ways about the subject matter of the survey) will choose not to respond/to respond; the groups of individuals who respond/don’t respond are the same w/ respect to the variable of interest & thus we’ve under/overrepresented a group in our population

wording bias

bias which occurs when the wording of the question(s) being asked influences the responses in a SYSTEMATIC way

what are some situations in which we might see response bias?

the respondent might not want to give a truthful answer to the question (e.g. income questions); the respondent might not UNDERSTAND the question (ambiguous wording); the respondent might want to please the interviewer; the order in which options were listed might influence the answers of the respondent; etc.

a difference between treatment results/treatment & no treatment results is statistically significant if…

this difference is too large to attribute to chance/natural random variation

experiment

a study in which researchers ASSIGN individuals to treatments/impose treatments on individuals

observational study

a study in which researchers observe individuals and record data but DON’T impose treatments

retrospective studies study what is in the ____; prospective studies will study events in the…

past; future

confounding variable

a variable that has an effect on the outcomes of the study but whose effects can’t be separated from those of the treatment variable

statistical control

a situation in which all variables other than the one of interest are HELD CONSTANT across all treatment groups

what are the 3 basic principles all experiments must incorporate?

randomization, replication, and control

randomization is done to…

equalize groups so that the effects of extraneous variables are equally distributed among groups

replication is done to…

minimize effects of random chance on our results (e.g. ensure differences between results for our groups aren’t due to random extremity)

control is enforced to…

minimize confounding; ensure groups only differ with respect to variable of interest

completely randomized design

subjects randomly assigned to treatment groups, treated; effects of treatments compared

placebo

an inactive treatment used when effects of treatment are compared to effects of no treatment (placebo only used for HUMAN subjects)

blinding

preventing a party in the experiment from knowing which treatment they received (single blind: either subject or evaluator doesn’t know; double blind: both subject and evaluator don’t know)

when describing a method of randomization, be sure to…

specify method of randomization, specify whether sampling w/ or w/out replacement, and (if using random digits table) specify what represents 1 [e.g. 01 is 1]

randomized block design

completely randomized experiment is performed WITHIN each block, results compared WITHIN blocks

matched-pairs design

two subjects similar to each other are paired, becoming a block, or one person becomes a pair & their before/after results are compared