What is the experimental unit?

• The smallest thing independently assigned to a treatment
• Every single measurement the machine records
• The largest room in the laboratory

What must a design name before data collection?

• The journal for submission
• The treatments, the units, and the response
• The color scheme for the figures

Random assignment balances even the factors nobody thought to name.

Circle one:   True   False

Patients come from several clinics with different baselines. What do you do?

• Block by clinic, then randomize treatments inside each block
• Drop every clinic except the largest one
• Assign all sick patients to the new drug

Twenty plants share one treated tray. What is the honest n for that dose?

• Twenty, one per plant
• One tray, averaged to a single value
• Forty, counting leaves as well

What role does a placebo arm play in a human study?

• It receives everything except the active factor, isolating the treatment gap
• It guarantees every patient is cured
• It replaces the need for random assignment

A trial randomized well but recruited only volunteers from one gym. What is threatened?

• Nothing, since randomization fixes everything
• Internal causality, which now collapses
• Generalization outside, since sampling was narrow

A reviewer attacks a study with n equals 12 from two cages. What is the likely flaw?

• The unit was miscounted: cages, not animals, were assigned
• Twelve is always too small for any study
• Reviewers may never question sample size