1. For discussion during lecture #6:
Problems from the Ott and Longnecker text: 4.77, 5.41, 11.65, 11.66, 11.30, 11.31 (Note: there is a typo, the question is supposed to refer to the test for the slope beta1, not the intercept beta0), 11.32. For the data from problem 11.65, replicate their results using a statistical package.
2. For discussion during lecture #11:
Problems from the Ott and Longnecker text (some require use of a computer package): 11.33, 11.34, 11.43, 11.52, 11.82, 11.83, 11.84, 11.85. Also for the small cereal data set (with lecture 4), use the matrix-based approach to regression to find the least-squares estimate of b0 in the model yi = b0 + ei, where yi is the calories in a serving of cereal.
3. For discussion during lecture #18:
Problems from the Ott and Longnecker text: 12.15, 12.16, 12.22, 12.23, 12.24, 12.28, 12.29, 12.56, 13.53 (verify their output with your own computer program).
4. For discussion during lecture #25:
Problems from the Ott and Longnecker text: 13.52, 13.67, 8.7, 8.29, 8.30, 8.34. Review the following SAS output, performing stepwise regression with a second-order model for the auto data. Any of the F tests for Entry or Removal can be written in the form F = [ SS(Regression,complete) - SS(Regression, reduced) ]/ MS(Residual, complete) . In step 7, under Statistics for Entry the printout shows the F for wt2 is 2.70. In step 8, under Statistics for Removal the printout shows the F for hpwt is 0.57. For each of these F statistics identify the complete and reduced models that are used.
5. For discussion during lecture #32:
Problems from the Ott and Longnecker text: 9.3, 9.14a,b (Calculate the Fisher LSD and Tukey W values yourself from the ANOVA table, check against their computer output), 15.6, 15.7, 15.10, 14.23, 14.24, 14.25. Also for the cuckoo data, construct a contrast to compare the average of the mean egg lengths for the two Pipit species to the average of the mean egg lengths for the other four species. Conduct a test for this contrast first as if it were a priori using a t test, and second as if it were a post hoc contrast using Scheffe's method.
6. For discussion during lecture #38:
For the data from problem 15.6, use Friedman's method to test for differences in performance according to background music, and compare your results to those from the RCB analysis of variance.
For these data do the following analyses: a) write a program to define dummy variables for the production line factor and enter the data, b) use a regression procedure (such as Proc Reg) to test the ANOVA null hypothesis of equality of the five groups, c) use the regression parameter estimates to calculate estimates of the group means. You can check your regression results with an ANOVA program if you wish.
For these data do the following analyses: a) Create a plot of x and y with the plotting symbol showing which process was used. Comment on the parallelism assumption and the likelihood of having group differences based on the plot. b) Define dummy variables and crossproduct terms. Use a regression procedure to test the parallelism assumption. c) If parallelism is ok, then proceed to test the equality of adjusted group means. d) Using the output from part c, calculate adjusted group means. e) Create residual-by-predicted and residual-by-hat-value plots to assess model assumptions. Do they appear ok? You can check your regression results with an ANCOVA program if you wish.