360 Chapter 11 Simple Analysis of Variance
Induced Mood
Negative Positive None
798
5125
16 7 11
13 3 9
13 10 11
24 4 10
20 5 11
10 4 10
11 7 7
795a. Run the appropriate analysis of variance and draw your own conclusion.
b. Which column means would you be interested in comparing for theoretical reasons when
we get to discussing multiple comparisons in the next chapter? Computer Exercises
11.28 In Exercise 7.46 you had data on students who had lost a parent through death, who came
from a divorced household, or who grew up with two parents. You then ran three separate t
tests comparing those groups.
a. Now reanalyze those data using an analysis of variance with GSIT as the dependent
variable.
b. How does your answer to this question differ from your answer in Chapter 7?
Use the following material to answer Exercises 11.29–11.31.
Introini-Collison and McGaugh (1986) examined the hypothesis that hormones normally pro-
duced in the body can play a role in memory. Specifically, they looked at the effect of post-
training injections of epinephrine on retention of a previously learned discrimination. To
oversimplify the experiment, they first trained mice to escape mild shock by choosing the left
arm of a Y maze. Immediately after training they injected the mice with either 0.0, 0.3, or
1.0 mg/kg of epinephrine. (The first group was actually injected with saline.) They predicted
that low doses of epinephrine would facilitate retention, whereas high doses would inhibit it.
Either 1 day, 1 week, or 1 month after original training, each mouse was again placed in the
Y maze, but this time was required to run to the right arm of the maze to escape shock. Pre-
sumably the stronger the memory of the original training, the more it would interfere with
the learning of this new task and the more errors the subject would make.
There are two data sets for this experiment, and they are described in Appendix Computer
Exercises. The original study used 18 animals in the three dosage groups tested after 1 day,
and 12 animals in each group tested after intervals of 1 week and 1 month. Hypothetical
data that closely reproduce the original results are contained in Epinuneq.dat, although for
our purposes there are data for only 7 subjects in the 1.0 mg/kg dose at the 1-month test.
A second data set was created with 12 observations in each of the 9 cells, and is called
Epineq.dat. In both cases the need to create data that were integers led to results that are slightly
conservative relative to the actual data, but the conclusions with respect to are the same.
11.29 On the reasonable assumption that there are no important differences from one interval to
the next, combine the data by ignoring the Interval variable and run the analysis of variance
on Dosage. Use the data in Epinuneq.dat. (You will have 42 observations for the 0.0 and
0.3 mg/kg doses and 37 subjects for the 1.0 mg/kg dose.)
11.30 Use the data in Epinuneq.dat to run three separate one-way analyses of variance, one at each
retention interval. In each case, test the null hypothesis that the three dosage means are
equal. Have your statistical package print out the means and standard deviations of the three
dosage groups for each analysis. Now run a separate analysis testing the hypotheses that the
three Interval means are equal. In this case you will simply ignore Dosage.
11.31 Rerun Exercise 11.29, this time using Epineq.dat. (The results will differ somewhat because
the data are different.) Calculate the average of the three error terms (MSerror) and show thatH 0