16.18 Compute the energy savings per household for the data in Exercise 16.16 by subtracting this
year’s bill from last year’s bill. Then run an analysis of variance on the savings scores and
compare that to the analysis of covariance.
16.19Klemchuk, Bond, and Howell (1990) examined role-taking in children. Children were
administered a battery of role-taking tasks. They were classified as being in daycare or not
being in daycare, and as ages 2–3 or ages 4–5. The hypothesis was that children with day-
care experience would perform better on role-taking tasks. The data are available at the
book’s Web site as Ex16-19.dat. Run the appropriate analysis.Computer Exercises
16.20 Use the data set named in Epinuneq.dat on the instructor’s disk to examine the results of the
study by Introini-Collison and McGaugh (1986) described prior to Exercises 11.28–11.31.
Using any statistical package, run a two-way analysis of variance with unequal sample
sizes. What would you conclude from this analysis?
16.21 Use the data from Mireault (1990) in the file named Mireault.dat referred to in Exercise 7.6
to run a two-way analysis of variance on the Global Symptom Index Tscore (GSIT) using
Gender and Group as independent variables. Plot out the cell means and interpret the results.
16.22 Using the same data as in Exercise 16.21, run an analysis of covariance instead, using year
in college (YearColl) as the covariate.
a. Why would we want to consider YearColl as a covariate?
b. How would you interpret the results?
16.23 In Exercise 16.22 we used YearColl as the covariate. Run an analysis of variance on
YearColl, using Gender and Group as the independent variables. What does this tell us that
is relevant to the preceding analysis of covariance?
16.24 Everitt reported data on a study of three treatments for anorexia in young girls. One treat-
ment was cognitive behavior therapy, a second was a control condition with no therapy, and
a third was a family therapy condition. These are the same data we examined in Chapter 14.
The data follow.626 Chapter 16 Analyses of Variance and Covariance as General Linear Models
Group Pretest Posttest Gain
1 80.5 82.2 1.7
1 84.9 85.6 0.7
1 81.5 81.4 2 0.1
1 82.6 81.9 2 0.7
1 79.9 76.4 2 3.5
1 88.7 103.6 14.9
1 94.9 98.4 3.5
1 76.3 93.4 17.1
1 81.0 73.4 2 7.6
1 80.5 82.1 1.6
1 85 96.7 11.7
1 89.2 95.3 6.1
1 81.3 82.4 1.1
1 76.5 72.5 2 4.0
1 70.0 90.9 20.9
1 80.4 71.3 2 9.1
1 83.3 85.4 2.1
1 83.0 81.6 2 1.4
1 87.7 89.1 1.4
1 84.2 83.9 2 0.3
1 86.4 82.7 2 3.7Group Pretest Posttest Gain
1 76.5 75.7 2 0.8
1 80.2 82.6 2.4
1 87.8 100.4 12.6
1 83.3 85.2 1.9
1 79.7 83.6 3.9
1 84.5 84.6 0.1
1 80.8 96.2 15.4
1 87.4 86.7 2 0.7
2 80.7 80.2 2 0.5
2 89.4 80.1 2 9.3
2 91.8 86.4 2 5.4
2 74.0 86.3 12.3
2 78.1 76.1 2 2.0
2 88.3 78.1 2 10.2
2 87.3 75.1 2 12.2
2 75.1 86.7 11.6
2 80.6 73.5 2 7.1
2 78.4 84.6 6.2
2 77.6 77.4 2 0.2
2 88.7 79.5 2 9.2
2 81.3 89.6 8.3