Use the preparation materials to prepare hand-written answers for the following questions. Please ask any questions on the "Questions -- Preparation Guide" on the course Team (see link on the homepage).
  1. Thoroughly compare and contrast a (generic) simple and full models. You should have at least five characteristics that you discuss.
  2. Define Yij, Ȳi., Ȳ.., ni, n, εij.
  3. What is the simple model in a 2-sample t-test. Express this in symbols, in words, and graphically.
  4. What is the full model in a 2-sample t-test. Express this both in symbols, in words, and graphically.
  5. What is SSTotal? Express this as a formula and in words with respect to residuals from a model (be specific about which model), as a measure of lack-of-fit, and graphically.
  6. What is SSwithin? Express this as a formula and in words with respect to residuals from a model (be specific about which model), as a measure of lack-of-fit, and graphically.
  7. Show how SSTotal partitions into two components.
  8. What is SSAmong? Express this as a formula and in words with respect to comparing two models (be specific about models), as it relates to lack-of-fit, and graphically.
  9. Explain how SSAmong changes as the difference in group means is increased.
  10. Explain how SSAmong changes relative to SSWithin as the difference in group means is increased.
  11. Is SSAmong a measure of “signal” or “noise”? A measure of “benefit” or “cost” (of using the full model)?
  12. What does dfAmong measure? Express this in relation to two models.
  13. How do you convert any SS to a “true variance”?
  14. MSTotal and MSWithin measure the variability of what “entities”?
  15. How do MSTotal and MSWithin differ?
  16. MSAmong measures the variability of what “entities”?
  17. How is an F test statistic computed?
  18. What does a “large” F test statistic mean? Answer this in terms of “signal” and “noise”, variability explained and unexplained (be specific about what is “explaining”), simple and full models, and null and alternative hypotheses.
  19. What does a “small” p-value (computed from an F test statistic) mean? Answer this in terms of “signal” and “noise”, variability explained and unexplained (be specific about what is “explaining”), simple and full models, and null and alternative hypotheses.