Statistics / Unit 5: Collecting Data
5.5 Designing Experiments
Notes
A well-designed experiment follows these principles.
- Control: compare treatments, usually with a control group (no treatment or a standard treatment). A placebo is a fake treatment that looks real.
- Random assignment: use chance to put subjects into treatment groups so the groups are alike before treatment.
- Replication: use enough subjects so differences are unlikely to be due to chance.
- Blinding: in a single-blind experiment subjects do not know their treatment. In a double-blind experiment neither the subjects nor those measuring the response know.
- Example: To test a new cold remedy, randomly assign 100 volunteers to the remedy or a placebo, with a double-blind design.
Practice
The purpose of random assignment in an experiment is to:
Chance balances other variables across groups, so differences in results can be credited to the treatment.
Why is a placebo group often included in a medical experiment?
A placebo shows how much improvement comes from expectation alone.
In a double-blind experiment:
Double-blinding prevents expectations from influencing the results.
Ready for more? Take the Unit 5 test.