Reference: Introduction to Statistical Reasoning, Gary Smith, "New Coke versus Old Coke" Example 4.6 page 186-187.
Concepts: Matched pairs experimental design, confounding variables
Materials: paper cups (2 per person), two flavors of heavily advertised soda pop (1 can will do about 16 people if you dispense 1-1/2 inch per cup)
Preparation: Read "New Coke versus Old Coke", prepare two identical containers for the two brands of soda. Mark one with a square and one with a circle.
1) Have each student write on a piece of scratch paper their stated preference for Coke or Pepsi (or Sprite v.s. 7-Up) with their names or a code number, and why they choose as they do. Collect the papers. Students should NOT state aloud their preferences.
2) Each student should take two paper cups. Mark one cup with a circle and one with a square on the bottom of the cup. Dispense a small amount of soda pop into the appropriate cups from your own larger dispensers with matching marks.
3) Have each student taste the soda. Vote silently for the one they prefer on a piece of scratch paper (indicate square or circle). Label with name or code number. Collect the votes.
4) Display tallies of the two votes: Stated Preference vs. Actual Preference and Circle vs. Square.
Calculate how many guessed correctly. Ask for comments about any disparity.
5) Read aloud "New Coke versus Old Coke" from Gary Smith's book.
6) Ask: Which is more important in determining market preference, stated preference or actual preference?
7) How big of a role does advertising play in the stated preference? Was advertising a confounding variable?
8) What type of experimental design was this? How the experiment be improved?
9) What is the survey design? How could you design a random sampling
procedure for our high school?