Selecting an Experimental Design - AP Statistics

Each participant receives multiple treatments, reducing variability. Each subject serves as their own control group.

Experiments involve the manipulation of variables; observational studies do not. Only experiments can establish cause-and-effect relationships.

Effects of a previous treatment that influence outcomes in subsequent treatments. Previous treatments may affect responses to later treatments.

The variable that is manipulated by the researcher. The factor deliberately changed to test its effect.

An experiment where two or more factors are tested simultaneously. Tests multiple factors and their interactions simultaneously.

Less effective if there is variability among experimental units. Cannot control for differences between experimental units.

To isolate the effect of the independent variable. Shows changes due to treatment, not other factors.

To increase reliability and accuracy of results. Multiple trials reduce impact of random variation.

When comparing two treatments with closely related subjects. Pairing controls for individual differences between subjects.

Randomized block design. Blocking controls known sources of variation.

A variable not of interest but could affect the outcome. Must be controlled to avoid confounding the results.

Subjects are randomly assigned to treatment and control groups. Gold standard design for establishing causal relationships.

To assess the consistency of interaction effects. Multiple replications verify interaction effect reliability.

To eliminate bias and balance unknown factors. Random assignment distributes unknown confounders equally across groups.

Subjects are divided into blocks based on a characteristic, then randomized within blocks. Controls for known sources of variation before randomization.

To serve as a baseline for comparison to the treatment group. Establishes what happens without the experimental treatment.

Replication. Multiple subjects per treatment reduce random variation effects.

Single-blind: participants unaware of treatment. Double-blind: both participants and experimenters unaware. Prevents bias from expectations affecting results.

To ensure each participant has an equal chance of receiving any treatment. Creates comparable groups to isolate treatment effects.

A study where researchers observe subjects without manipulating variables. Cannot establish causation, only associations between variables.

An experimental design where subjects are randomly assigned to treatments. Ensures each subject has equal probability of any treatment assignment.

To serve as a baseline for comparison to the treatment group. Establishes what happens without the experimental treatment.

Less effective if there is variability among experimental units. Cannot control for differences between experimental units.

When the effect of one factor depends on the level of another factor. Combined factor effects differ from individual effects.

Crossover design. Each subject receives all treatments as their control.

To ensure each participant has an equal chance of receiving any treatment. Creates comparable groups to isolate treatment effects.

A design where subjects are paired based on similarities and given different treatments. Reduces variability by controlling for subject characteristics.

Subjects change their behavior because they know they are being studied. Awareness of observation changes natural behavior patterns.

Single-blind: participants unaware of treatment. Double-blind: both participants and experimenters unaware. Prevents bias from expectations affecting results.