A researcher considers two designs to compare a new and old textbook. Design 1: randomly assign 40 students to each book. Design 2: pair up students by GPA and randomly assign one in each pair to each book.
Why might Design 2 be preferred when GPA strongly predicts test scores?
- A
Design 1 has fewer experimental units
- B
Design 2 eliminates the need for randomization
- Ccheck_circle
Design 2 (matched pairs) controls for the variability due to GPA
- D
Design 1 cannot use random assignment
Explanation
Pairing on GPA removes its variability from the comparison, making it easier to detect a true textbook effect. Matched pairs is essentially blocking with pairs.