AP Statistics: Sampling and Experimental Design

Vocabulary practice flashcards covering statistical sampling methods, sources of bias, and experimental design principles.

Population

The entire group you wanna learn about.

Sample

The smaller group you actually collect data from.

Bias

A flawed method that makes your results consistently wrong in one direction.

Unbiased

A fair method that hits the true answer on average over the long run.

Simple random sample

Giving every possible group an equal chance to be picked.

Stratified random sample

Break the population into groups by a trait, then take a simple random sample from every group.

Block design

Group similar subjects together first (like by age), then randomly give treatments inside each group.

Strata

Groups divided by a trait used in the context of sampling ($\text{strata} \rightarrow \text{sampling}$).

Blocks

Groups of similar subjects assembled in the context of experiments ($\text{blocks} \rightarrow \text{experiments}$).

Block vs. strata

The methodological rule that strata applies to sampling, whereas blocks applies to experiments ($\text{strata} \rightarrow \text{sampling}$, $\text{blocks} \rightarrow \text{experiments}$).

Single-blind experiment

An experiment where the subjects do not know which treatment they are getting.

Double-blind experiment

An experiment where neither the subjects nor the evaluators know who got which treatment.

Evaluators (in a double-blind experiment)

The individuals who, along with the subjects, do not know who received which treatment.

Placebo

A fake, dummy treatment used as a baseline comparison.

Baseline comparison

The role served by a placebo (a fake, dummy treatment) in an experimental setup.

Matched pairs design/experiment

An experiment that compares two treatments by using highly similar pairs of people.

Confounded, nonrandomized study

A flawed study where treatments aren’t randomly assigned, making results mixed up with outside variables.

Outside variables

Factors kept constant by experimental controls across all groups so they don’t mess up results.

Cluster sampling

Splitting the population into diverse groups, randomly picking a few entire groups, and surveying everyone inside them.

Systematic random sampling

Picking a random starting point on a list, then surveying every $k^{\text{th}}$ person ($10^{\text{th}}$).

Every $k^{\text{th}}$ person

The fixed interval (such as every $10^{\text{th}}$ person) surveyed from a list after choosing a random starting point in systematic random sampling.

Convenience sample

A heavily biased method that only picks people who are easiest to reach.

Voluntary response sample

A heavily biased method where people volunteer themselves to join a poll (attracting people with strong opinions).

Random sampling

Using a chance process to pick people fairly, allowing you to generalize to the whole population.

Nonresponse

A situation where people are chosen but cannot be contacted or refuse to participate.

Undercoverage

A flaw where some members of the population are completely excluded from choosing the sample.

Question wording

Confusing or misleading language that tilts answers.

Response bias

A bias that occurs when people give inaccurate or dishonest answers.

Response variable

The outcome you measure at the end.

Observational study

A study where you watch and measure without interfering.

Experiment

A study where you intentionally impose a condition (treatment) to see what happens.

Random assignment

Using chance to place units into treatment groups.

Controls (in designing experiments)

Keeping outside variables constant for all experimental groups so they don’t mess up results.

Randomized block design

Group similar subjects together first (blocks), then randomize treatments inside those blocks.

Completely randomized design

Throw everyone’s name in one hat and assign treatments purely by chance.

Treatments

The specific conditions given to the units.

Experimental units

The smallest thing or group that receives a treatment in an experiment.

Principles of experimental design

The fundamental principles of experimental design designated by the acronym CCRR: Comparison, Control, Random assignment, and Replication.

CCRR

The acronym for the four principles of experimental design: Comparison, Control, Random assignment, Replication.

Comparison (principle of experimental design)

Compare $2$ or more groups.

Control (principle of experimental design)

Keep outside variables constant.

Random assignment (principle of experimental design)

Use chance to mix up groups.

Replication (principle of experimental design)

Use enough subjects to ensure it isn’t a fluke.

Condition to claim causation

Can only claim causation when treatments were randomly assigned.

Condition to generalize

Can only generalize when the subjects were randomly selected from the population.