AP Stats Experimental Design 3 — Questions and Answers
Question 1: An experiment uses a matched pairs design where each participant is tested under both the old and new software interfaces on different days. The order of interfaces is randomized for each participant. What does randomizing the order control for?
- Individual differences between participants
- Carryover or order effects (Correct answer)
- Differences in software difficulty
- Sample size adequacy
Correct answer: Carryover or order effects
Randomizing order in a matched pairs design controls for carryover and order effects that could bias results.
Question 2: A clinical trial randomly assigns 500 patients to drug A and 500 to a placebo. After the trial, it is discovered that patients in the drug A group were, on average, younger. This is an example of:
- Blocking failure
- Chance imbalance due to random variation (Correct answer)
- Deliberate confounding
- Non-response bias
Correct answer: Chance imbalance due to random variation
Random assignment does not guarantee perfect balance — chance variation can create group differences, especially with smaller samples.
Question 3: Which of the following best describes the principle of replication in experimental design?
- Repeating the entire experiment multiple times in different labs
- Using enough experimental units per treatment to detect real effects (Correct answer)
- Assigning each subject to every treatment condition
- Measuring the response variable more than once per subject
Correct answer: Using enough experimental units per treatment to detect real effects
Replication means having multiple experimental units per treatment group so that results are reliable and not due to a single unusual unit.
Question 4: A study compares two teaching methods by having Teacher A use Method 1 and Teacher B use Method 2. The teaching method is confounded with:
- The response variable (test scores)
- The teacher (a potential lurking variable) (Correct answer)
- The students' prior knowledge
- The time of year the study is conducted
Correct answer: The teacher (a potential lurking variable)
Because each teacher uses only one method, any observed difference could be due to the teaching method or the individual teacher.
Question 5: In an experiment testing three doses of a medication, a researcher uses a Latin square design to control for both time period and patient order effects. This is an extension of:
- Simple random sampling
- Blocking in two directions simultaneously (Correct answer)
- Completely randomized design principles
- Stratified random sampling
Correct answer: Blocking in two directions simultaneously
A Latin square design blocks in two directions (rows and columns) simultaneously, controlling for two sources of variation.
Question 6: A researcher claims that drinking coffee causes increased productivity because a survey shows coffee drinkers are more productive. The major flaw in this causal claim is:
- The sample size is too small
- The study is observational and cannot establish causation (Correct answer)
- Coffee drinkers were not blinded to their treatment
- Productivity was not measured accurately
Correct answer: The study is observational and cannot establish causation
A survey is observational — without random assignment to treatments, causal claims cannot be supported.
Question 7: When comparing a new exercise program to a control group, a researcher finds that both groups improve over time. The improvement in the control group is most likely due to:
- The placebo effect
- Regression to the mean and natural improvement over time (Correct answer)
- Confounding from the treatment group
- Measurement error in fitness testing
Correct answer: Regression to the mean and natural improvement over time
Control groups often show improvement due to natural changes over time, regression to the mean, and general health awareness effects.
AP Stats Experimental Design 3
An experiment uses a matched pairs design where each participant is tested under both the old and new software interfaces on different days.
The order of interfaces is randomized for each participant.
What does randomizing the order control for?