One-

One-Way
Analysis Of Variance

Brawijaya Professional Statistical Analysis

BPSA MALANG
Jl. Kertoasri 66 Malang
(0341) 580342
One-Way Analysis of Variance

You can use the One-Way ANOVA procedure to test the hypothesis that the means of
two or more groups are not significantly different.

One-Way ANOVA also offers:

• Group-level statistics for the dependent variable
• A test of variance equality
• A plot of group means
• Range tests, pairwise multiple comparisons, and contrasts, to describe the nature of
the group differences

Testing the Equality of Group Variances

An important first step in the analysis of variance is establishing the validity of

assumptions. One assumption of ANOVA is that the variances of the groups are
equivalent. This example illustrates how that test is performed.

A sales manager wishes to determine the optimal number of product training days
needed for new employees. He has performance scores for three groups: employees with
one, two, or three days of training. The data are in the file salesperformance.sav .

An Error Bar Chart

Before running the analysis of variance, you graph the means and standard errors.
To create an error bar chart, from the menus choose:

Graphs
Error Bar...

1 - One-Way Analysis of Variance

Click Define.

Select Score on training exam as the analysis variable.

Select Sales training group as the category variable.

2 - One-Way Analysis of Variance

 Select Standard error of mean from the Bars Represent drop-down list.
Click OK.

Average performance clearly increases with added training days, but variation in
performance decreases at the same time. ANOVA assumes equality of variance across
groups; that assumption may not hold for these data.

Running the Analysis

To test the equality of variance assumption, from the menus choose:

Analyze
Compare Means
One-Way ANOVA...

3 - One-Way Analysis of Variance

 Select Score on training exam as the dependent variable.
Select Sales training group as the factor variable.
Click Options.

Select Descriptive and Homogeneity of variance test.

Click Continue.
Click OK in the One-Way ANOVA dialog box.

Descriptive Statistics Table

The standard deviation and standard error statistics confirm that as training days
increase, variation in performance decreases.

4 - One-Way Analysis of Variance

Levene Test Table

The Levene statistic rejects the null hypothesis that the group variances are equal.
ANOVA is robust to this violation when the groups are of equal or near equal size;
however, you may choose to transform the data or perform a nonparametric test that
does not require this assumption.

Performing a One-Way ANOVA

In response to customer requests, an electronics firm is developing a new DVD player.

Using a prototype, the marketing team has collected focus group data. ANOVA is being
used to discover if consumers of various ages rated the design differently.

This example uses the file dvdplayer.sav .

Running the Analysis

To begin the analysis, from the menus choose:

Analyze
Compare Means
One-Way ANOVA...

5 - One-Way Analysis of Variance

 Select Total DVD assessment as the dependent variable.
Select Age Group as the factor variable.
Click Options.

The means plot is a useful way to visualize the group differences.

Select Means plot.
Click Continue.
Click OK in the One-Way ANOVA dialog box.

ANOVA Table

The significance value of the F test in the ANOVA table is 0.001. Thus, you must reject
the hypothesis that average assessment scores are equal across age groups. Now that
you know the groups differ in some way, you need to learn more about the structure of
the differences.

6 - One-Way Analysis of Variance

 A Plot of Group Means

The means plot helps you to "see" this

structure. Participants between the ages
of 32 and 45 rated the DVD player more
highly than their counterparts. If more
detailed analyses are desired, then the
team can use the range tests, pairwise
comparisons, or contrast features in
One-Way ANOVA.

Contrasts between Means

In general, F statistics establish that there is or is not a difference between group means,
and means plots suggest where the difference may lie. As this section shows, you can
use the One-Way ANOVA procedure to specify exactly how the means differ and test
those specifications.

Looking at the DVD data, market researchers ask:

• Are the two groups between the ages of 32 and 45 really different?
• Can participants under 32 and over 45 be considered statistically equivalent?

This example uses the file dvdplayer.sav .

Running the Analysis

To begin the analysis, from the menus choose:

Analyze
Compare Means
One-Way ANOVA...

7 - One-Way Analysis of Variance

 Select Total DVD assessment as the dependent variable.
Select Age Group as the factor variable.
Click Contrasts.

The first contrast compares only groups 3 and 4; the others are eliminated by giving them
weights of 0.

Type 0 as the coefficient for group 1, then click Add.

Add another coefficient of 0 for group 2.

Next, add the coefficients for groups 3 and 4 so that together they sum to 0.
Type -1 as the coefficient for group 3, then click Add.

8 - One-Way Analysis of Variance

 Add a coefficient of 1 for group 4.
Type 0 as the coefficient for group 5,
then click Add.
Add another coefficient of 0 for group
6.
Click Next to move to the second
contrast.

The second contrast compares groups 1

and 2 with groups 5 and 6. Groups 3
and 4 are eliminated.
Type .5 as the coefficient for group 1,
then click Add.

Add another coefficient of .5 for

group 2.
Next, groups 3 and 4 are each
given coefficients of 0.
Type 0 as the coefficient for
group 3, then click Add.
Add another coefficient of 0 for
group 4.

So the sum of contrast coefficients will

equal 0, groups 5 and 6 receive
coefficients of -.5.
Type -.5 as the coefficient for group 5,
then click Add.
Add another coefficient of -.5 for
group 6

9 - One-Way Analysis of Variance

 Click Continue.
Click OK in the One-Way ANOVA dialog box.

Contrast Coefficients Table

The contrast-coefficient table is a convenient way to check that the proper weights were
given to the groups.
If the mean assessments of the DVD player for the 32-38 and 39-45 age groups are
equal, then you would expect the observed difference in the mean assessment for these
groups to be near 0. By specifying -1 and 1 as the contrast coefficients for these groups,
the first contrast tests whether the observed difference is statistically significant.
• Similarly, if the mean assessments of the under-32 age groups and over-45 age
groups are equal, you would expect the sum of the first two groups to be equal to the
sum of the last two groups, and the difference of these sums to be near 0.

10 - One-Way Analysis of Variance

Contrast Test Table

Note that the results are displayed in two panels: the first assumes that the variances of
the groups are equal, and the second assumes that they are unequal. In this case, the
variances of the groups are assumed equal, so we focus on the first panel.
• The significance values for the tests of the first contrast are both larger than 0.10.
This indicates that the age 39-45 group is not significantly more favorable toward the
DVD player than the age 32-38 group.
• Likewise, the significance values for the tests of the second contrast are larger than
0.10. Participants under 32 and over 45 have statistically equivalent assessment
scores.

All Possible Comparisons between Means

Contrasts are an efficient, powerful method for comparing exactly the groups that you
want to compare, using whatever contrast weights that you require. However, there are
times when you do not have, or do not need, such specific comparisons. The One-Way
ANOVA procedure allows you to compare every group mean against every other, a
method known as pairwise multiple comparisons.

A sales manager has analyzed training data using One-Way ANOVA. While significant
group differences were found, there are no prior hypotheses about how the three groups
should differ. Therefore, he has decided to simply compare every group to every other
group.

This example uses the file salesperformance.sav .

11 - One-Way Analysis of Variance

Running the Analysis

To begin the analysis, from the menus choose:

Analyze
Compare Means
One-Way ANOVA...

Select Score on training exam as the dependent variable.

Select Sales training group as the factor variable.
Click Post Hoc.

12 - One-Way Analysis of Variance

The Post Hoc tests are divided into two sets:
• The first set assumes groups with equal variances.
• The second set does not assume that the variances are equal. Because the Levene
test has already established that the variances across training groups are
significantly different, we select from this list.
Select Tamhane's T2.
Click Continue.
Click OK in the One-Way ANOVA dialog box.

Post Hoc Test Table

• The group with one training day performed significantly lower than the other groups.
• Trainees with two and three days do not statistically differ in average performance.
Despite this equality, the manager may still consider the added benefit of the third
training day, given the large decrease in variability.

13 - One-Way Analysis of Variance

Robust Analysis of Variance
Post hoc results are valid to the extent that the standard F statistic is robust to violations
of assumptions. As mentioned before, the F statistic is robust to unequal variances when
sample sizes are equal or nearly equal. However, when both the variances and the
sample sizes differ, the standard F statistic lacks power and is prone to give incorrect
results. This section discusses two analysis of variance methods available in the One-
Way ANOVA procedure that provide an alternative in these circumstances.

The management of a local bank has received complaints about the amount of time that
customers spend waiting in line at one of their facilities. In response, analysts have
recorded wait times at that facility and two other area banks for comparison purposes.

This example uses the file waittime.sav .

Before the Analysis

A boxplot shows that branch A, the bank receiving the complaints, contains outliers.
ANOVA is robust to unequal variances with groups of near equal size; however, there are
an unequal number of observations per branch. In this case, they can use the One-Way
ANOVA procedure to obtain robust F statistics.

14 - One-Way Analysis of Variance

Running the Analysis

To begin the analysis, from the menus choose:

Analyze
Compare Means
One-Way ANOVA...

Select Wait time in minutes as the dependent variable.

Select Bank branch as the factor variable.
Click Options.

15 - One-Way Analysis of Variance

 Select Homogeneity of variance test, Brown-Forsythe, and Welch.
Click Continue.
Click OK in the One-Way ANOVA dialog box.

Levene Test Table

The Levene test confirms the suspicion that the variances of the groups are different.

Standard ANOVA Table

The p value associated with the standard ANOVA F statistic is very close to .05.
However, because the variances and the group sizes are unequal, we are uncertain
whether to trust these results.

16 - One-Way Analysis of Variance

Robust Test Table

• Contrary to the finding for the standard F statistic, the Brown-Forsythe test statistic is
significant below .05.
As with the standard F statistic, the Welch statistic is not significant below .05. The Welch
statistic is more powerful than the standard F or Brown-Forsythe statistics when sample
sizes and variances are unequal. The biasing effect of the outliers can be assessed by
removing them and rerunning the ANOVA.

Summary

With the One-Way ANOVA procedure, you are able to:

• Validate the assumption of variance equality
• Obtain the ANOVA table and results
• Visually inspect the group means
• Perform custom contrasts, tailored to your specific hypotheses
• Compare each mean to every other mean, assuming variance equality or not
• Perform two types of robust analysis of variance

Related Procedures

The One-Way ANOVA procedure is used to test the hypothesis that several sample
means are equal.
• You can alternately use the Means procedure to obtain a one-way analysis of
variance with a test for linearity.
• If your test variable does not meet the conditions of analysis of variance, you can
use the Kruskal-Wallis H test in the Tests for Several Independent Samples
procedure.
• If you have multiple factors, use the GLM Univariate procedure to obtain two-way
analysis of variance, analysis of covariance, and more.

17 - One-Way Analysis of Variance

Recommended Readings

Brown, M. B., and A. B. Forsythe. 1974. Robust tests for the equality of
variances. Journal of the American Statistical Association, 69:, 264-267

Levene, H. 1960. . In: Contributions to Probability and Statistics, I. Olkin, eds.

Palo Alto, CA: Stanford University Press.

Milliken, G., and D. Johnson. 1992. Analysis of Messy Data: Volume 1. Designed
Experiments. New York: Chapman & Hall.

Neter, J., W. Wasserman, and M. H. Kutner. 1990. Applied Linear Statistical

Models. Homewood, IL: Irwin.

Welch, B. L. 1951. On the Comparison of Several Mean Values: An Alternative

Approach. Biometrika, 38:, 330-336.

18 - One-Way Analysis of Variance