AP Statistics Unit 3 Review

Confounding variable

Any variable that biases the results and/or conclusion of an experiment. Lurking variables could be an example of this.

Convenience Sample

A sample that is easy to get (produces bias)

Census Survey

a survey that covers the whole population

Voluntary Response Sample

When people chose to answer

Simple Random Sample (SRS)

Every member of the population has a known and equal chance of selection

Stratified Random Sample

The population is divided into smaller groups based on unifying characteristics in order to reduce variability, after which SRS is performed within each strata.

Cluster Sample

Samples one group from a population of groups, assuming that each group is representative of the population. (eg. picking one box of toys from an entire shipment of toys, assuming all boxes are representative of each other and just surveying one)

Systematic Sample

From the sampling frame, every nth sample is surveyed.

Experiments

Experiments compare treatments, have random assignment, have replication, and have control.

Observational Studies

No treatments were imposed

What can you conclude from an observational study?

Association within a population

What can you conclude from an experiment that is well-designed?

Causation within a population

A sample of students from CPHS shows a result to be statistically significant. What is the population that the results of this experiment can be extrapolated to?

All students within CPHS

Random integers on the calculator

1: Seeding

type any large #

hit sto->math PROB: 1:rand

2: To Get #'s

math PROB 8:randIntNoRep (lower, upper, n)

Undercoverage Bias

Some groups are left out of the sampling process

Nonresponse Bias

individual who was chosen decides not to cooperate

Voluntary Response Bias

Invitation is extended to all. Those who choose to participate may differ from those who don't.

Response Bias

When the behavior of the respondent or interviewer causes bias.

When Making an Experiment Diagram

- ALWAYS show random assignment

- Start with x subjects, arrow to random assignment, split to equal groups and specify number of subjects in each, arrow to treatments for each group, then return all branches of the diagram to a statement saying to compare the measured variable

When Writing About Bias

- identify population and sample

- explain how sample may differ from population

- identify bias

- explain how bias may lead to over or under estimate

When Designing an Experiment

- state what type of experiment

- state number of treatments and experimental units

- state how each treatment group is selected through random assignment

- state blinding and placebo

- compare response variable

Single Blind

Subjects don't know their treatment

Double Blind

Subjects and surveyors don't know their treatment

Placebo

People can get better just thinking they will

Blocking

Allows us to control confounding variables by making sure they are evenly spread to all treatment groups

Randomized Block Design

- separate subjects into blocks

- randomly assign treatments within each block

Matched Pairs Design

Like subjects are paired and each is randomly assigned to a treatment

Statistical Inference

If experimental units are representative of the population, then the results can be generalized to the population

Sampling Variability

Different samples yield different estimates. Larger samples produce more accurate results

Margin of Error

Creates an interval of plausible values (will learn later but I think it's 5%)

Statistically Significant

Observed changes between treatment groups that are larger than what you'd get by chance alone make the difference likely to be real.