Professional Documents
Culture Documents
NOTE TO INSTRUCTORS
Chapter 10 starts with defining all the new terms that are used in analysis of variance.
Most students will understand them fairly quickly. The one term they frequently have
trouble with is the “level” of the independent variable. I have provided a classroom
activity that allows them to practice labeling the levels of an independent variable. One-
way ANOVA is used instead of multiple t tests to control for experiment error. Students
usually understand how the probability of a Type I error adds up with each additional test.
It is very important to spend a fair amount of class time explaining between-group
variability and within-group variability and how the two types of variability contribute to
the F ratio. There is a PowerPoint template that shows students an example of two
different experiments. Both experiments have three groups and the same differences
between the groups but the within-groups variability is much greater for one experiment
compared to the other. The textbook also includes a useful graphic in Figure 10.4 (p.
335) that illustrates two examples of how total variance is split between within-group
variability and between-group variability. Since the F ratio is a ratio based on sums of
squared scores, all F values are positive and the distribution looks very different. Some
students have trouble looking up values in the more complex F table. Give the students
lots of practice in class so they get immediate feedback on whether they are using the
table correctly. You also need to go slowly over the three different degrees of freedom
for this test. I have provided both PowerPoint templates and handouts for showing
students step by step how to do the calculations. There is also a PowerPoint template
for the ANOVA summary table and a handout that allows students to practice filling out
the summary table based on small pieces of information. Students will again be using r2
as a measure of effect size, but there is a new formula for computing this measure. The
post-hoc test is fairly straightforward except for the use of the q value. I have provided
examples that result in some group differences showing a significant post-hoc test and
others not being significant based on the post-hoc test. Students should practice
interpreting these results.
OUTLINE OF RESOURCES
5. Handouts
• Handout 10-1: Identifying levels of an independent variable (p. 106)
• Handout 10-2: Filling in source tables (p. 107)
• Handout 10-3: Effect size and post-hoc tests (pp. 108-109)
6. PowerPoint templates
• PowerPoint template 10-1: Understanding between-groups and within-groups
variance (p. 110)
• PowerPoint template 10-2: The relationships among the values in a source
table (p. 111)
CHAPTER GUIDE
c) The ANOVA tells you that at least one pair of means is statistically different
from another, but it does not tell which pair or if more than one pair are
significant.
Chapter 10 Between-Subjects, One-Way Analysis of 100
d) A post-hoc test is a test conducted “after the fact,” i.e., as a follow-up, and
tells you exactly which pairs of means are significantly different.
f) Interpreting F ratios
When there is no treatment effect, the F ratio will be 1.
The larger the value of the F ratio is, the more likely the results are
significant. There cannot be a negative F ratio because it is a ratio of
squared scores.
The goal of this assignment is to help students practice identifying the levels of an
independent variable. Handout 10-1 has several examples. Students should work in
pairs to name the independent variable and then name the levels. When they are done,
have students share their answers with the class.
The correct answers are: 1) The independent variable is stress; the levels are physical,
emotional, none. 2) The independent variable is color; the levels are red, brown, orange.
3) The independent variable is music; the levels are none, heavy metal, classical, pop.
4) The independent variable is color; the levels are pink, blue, green, white. 5) The
independent variable is caffeine; the levels are decaf coffee, half-caff coffee, regular
coffee. 6) The independent variable is breakfast; the levels are healthy, minimal, none.
b) The null hypothesis is H0: µ1 = µ2 = ... µκ. The alternative hypothesis is H1:
At least one population mean is different from the others.
c) The decision rule is if F > Fcv , then reject the null. If F < Fcv, fail to reject the
null.
d) There are three different degrees of freedom to calculate for this test.
1) The degrees of freedom for between-group is k – 1. This is the
degrees of freedom for the numerator (the difference between the
groups). It is the number of groups minus one.
2) The degrees of freedom for within-group is N – k. This is the degrees
of freedom for the denominators (within-groups variance). It is the
total number of people in the experiment (N) minus the number of
groups (k).
Chapter 10 Between-Subjects, One-Way Analysis of 102
f) Once you have calculated the three sums of squares, you can do a math
check by adding the SSBetween + SSWithin = SSTotal. If this does not add up,
then go back and check your math for an error.
h) Students can use the source table to calculate the mean square between
(MSBetween) and the mean square within (MSWithin).
MSBetween = SSBetween/dfBetween
MSWithin = SSWithin/dfWithin
j) The obtained F is then compared to the tabled value in Table 4a for α = .05
and Table 4b for α = .01. Students need to look up the F value by looking
Chapter 10 Between-Subjects, One-Way Analysis of 103
for the degrees of freedom in the numerator (dfBetween) along the top of the
chart and the degrees of freedom for the denominator (dfWithin) along the left
side of the chart.
a) r2 is used to calculate the effect size. This measure uses the SSBetween and
the SSTotal from the source table. The SSBetween is divided by the SSTotal and
then multiplied by 100 to get a percentage. This tells you the percentage of
variability in the scores that is accounted for by the treatment effect.
1) The closer r2 is to 100%, the stronger is the effect of the independent
variable.
2) The closer r2 is to 0%, the weaker is the effect of the independent
variable.
e) The Tukey HSD test was designed for experiments with the same number of
cases in each group. However, using the sample size from the smallest
group is acceptable, and this just makes the test more conservative.
f) Use the obtained means for each group to determine the direction of the
difference.
examples for students to work on. Students can work alone or in pairs and then share
their answers with the class.
HANDOUT 10-1
Identifying Levels of an Independent Variable
For each of the following research projects, please indicate the independent and
dependent variable and then specify the levels of the independent variable.
1. You want to know how different types of stress influence exam taking. You have 30
statistics students. Ten take their final exams after running on a treadmill for 15
minutes (physical stress), 10 after being told that they were doing poorly in the class
(emotional stress), and 10 after listening to soothing music (no stress). You compare
the final exam scores of each group.
2. You want to know how color influences taste perception. Each group of participants
is given colored “tea,” which is actually just colored plain water. One group has cups
with red fluid, a second group has brown fluid, and the third group has orange fluid.
The participants are to describe the flavor of each tea and rate how well they liked
each tea.
3. You want to know if music can influence heart rate. Some participants sit in a quiet
room, others hear loud heavy-metal music, while others hear soothing classical
music, and another group hears upbeat pop music. All participants have their heart
rate checked after five minutes.
4. You want to know if color can influence anxiety levels. One group of participants sits
in a pink classroom, a second group sits in a blue classroom, a third group sits in a
green classroom, and the final group sits in a white classroom. All participants rate
their anxiety on a 7-point scale where higher numbers mean more anxiety.
5. Ryan, Hatfield, and Hofstetter investigated the effects of caffeine on memory in older
adults. All participants were between the ages of 70 and 80. One group of
participants was given decaf coffee, one group was given half-caf coffee, and one
group was given full-strength caffeinated coffee. After finishing their coffee,
participants were given 16 words to memorize. They then took both a free recall and
a recognition test on the words.
6. There have been many studies showing that breakfast is the most important meal of
the day for children. You are interested in knowing if breakfast is also important for
college students’ learning. One-third of the college students are told to eat the
standard for a healthy breakfast (one serving each of fruit, cereal, bread, and a
protein); one-third of the students are told to eat a minimal breakfast (coffee and a
bagel), and one-third are told to eat nothing. All students must get up at 8:00 A.M.
and have finished their breakfast (or not eaten) by 8:45 A.M. At 11:00 A.M. all the
participants report to the laboratory and take a reading comprehension test.
Chapter 10 Between-Subjects, One-Way Analysis of 107
HANDOUT 10-2
Filling in Source Tables
Use what you know about the relationships among the values in a source table to fill in
the missing information. Then test to see if the F is significant.
1.
Source SS df MS F
between 12 3
within
total 72 33
2. . Source SS df MS F
between 10
within 120
total 140 42
3.
Source
SS df MS F
between 48 4
within 30 6
total
4. Use the following information to create a source table. There were 154 participants
divided into 4 groups. The F ratio is 15, and the mean square between is 30.
Chapter 10 Between-Subjects, One-Way Analysis of 108
HANDOUT 10-3
Effect Size and Post-Hoc Tests
For each of the following studies, use the information given to calculate the effect size;
then do post-hoc tests and interpret the results.
Source SS df MS F
between 48 2 24 4
within 72 12 6
total 120 14
2. You think that people work faster in the cold. To investigate this you have people do
a word search task in one of four rooms. You record how many minutes it takes to
complete the task. The room temperatures are set at 50, 60, 70, and 80 degrees.
The average time it takes to complete the task for the 50-degree room is 6 minutes,
for the 60-degree room is 2.3 minutes, for the 70-degree room is 3.7 minutes, and for
the 80-degree room is 6.7 minutes. There were three people in each group.
Source SS df MS F
between 36 3 12 6
within 18 9 2
total 54 12
Chapter 10 Between-Subjects, One-Way Analysis of 109
Source SS df MS F
Between 48 3 16 4
Within 48 12 4
Total 96 15
Chapter 10 Between-Subjects, One-Way Analysis of 110
Experiment A Experiment B
8 17 22 18 20 30
9 15 19 6 25 25
11 13 21 10 15 10
12 16 18 2 5 15
10 14 20 4 10 20
M 10 15 20 10 15 20
SD 1.6 1.6 1.6 7.1 7.9 7.9
Chapter 10 Between-Subjects, One-Way Analysis of 111
Source SS df MS F
MSb = SSb/dfb
MSw = SSw/dfw
F = MSb/MSw
Source SS df MS F