1. For discussion during lecture ~#9:
From the text: Exercise 2.3, Exercise 3.3, Exercise 4.4, Exercise 5.5, Problem 6.1, Exercise 7.2
2. For discussion during lecture # ~20:
From the text: From the text: Problem 8.3, Problem 10.1,
Exercise 11.4.
For the 2x2 small unbalanced factorial design discussed in our unbalanced
factorial data lecture, calculate the least squares mean and its standard
error for level 1 of factor B.
Problem 12.2, Exercise 13.2. For Exercise 13.2 calculate the efficiency of
using an RCB design.
Use the EMS information from the link with lecture ~16. For the 3 factor
model with one effect fixed and two random, specify the numerator and
denominator mean squares for the F test for the following effects: i) ABC
ii) AB and iii) A.
3. For discussion during lecture ~#34 :
From the text: Exercise 13.3 (use just the first four columns so it is an ordinary Latin square), Exercise 14.4, Problem 16.8, Exercise 15.2, Problem 15.1b, Problem 15.4.
Analyze the data from Problem 16.8 as if the split-plot factor were a
repeated measure. Use the downweighted split-plot degrees of freedom
method, the method of modeling covariance matrices, and the random
coefficient method. In SAS the first method requires the data in
multivariate format, which is available here.
4. For discussion during lecture ~#41:
i) Analyze this data set from Roger Kirk's experimental design text to determine which factors affect the response. List the alias structure and summarize your results.ii)
Construct a 25-2 fractional
factorial design. Use I=ABD=-CDE as the defining relation. What is the
resolution of the design? Write out the alias structure for main effects
and two-way interactions.
iii)
Problem 18.6. (Find an embedded 24 factorial
design. Find the 15 terms in the defining relation, then refer to the
aliases when interpreting your results.)
iv) For the linked data file, create a scatterplot of log sales and log assets, using symbols or colors to label the sectors that data are from. Then conduct an analysis of covariance to test for different in log sales by sector after adjusting for log assets. Calculate the adjusted means for log sales by sector, and calculate the adjusted mean of log sales for the HiTech by hand to check the adjusted mean calculation. Test the parallelism hypothesis for the ANCOVA. Assess the normality and constant variance assumptions for the ANCOVA model.
v) Problems 18.1a, 18.2d, and 18.2h