WEBVTT

1
00:00:03.033 --> 00:00:11.199
In this video, you will learn to perform
a one-way ANOVA using the
One-Way ANOVA task in SAS Studio.

2
00:00:11.200 --> 00:00:17.133
Let’s say that we want to determine
whether age is related to sense of smell.

3
00:00:17.133 --> 00:00:27.666
We can address this question by using a one-
way ANOVA to predict a continuous measure of 
sense of smell from a five-category age variable.

4
00:00:27.666 --> 00:00:38.066
First, click on Tasks in the panel on
the left side of the window. The bottom
category of tasks is labeled Statistics.

5
00:00:38.066 --> 00:00:48.232
If you do not already see the contents
of the statistics category, double-click
on Statistics to expand the category.

6
00:00:48.233 --> 00:00:58.133
Next, double-click on One-Way ANOVA.
The task menu appears to the right.

7
00:00:58.133 --> 00:01:03.733
Now that we have selected a task,
we do not need to keep the pane
on the left-hand side open.

8
00:01:03.733 --> 00:01:03.766
Click on the Maximize View button
to close the Tasks pane.

9
00:01:03.766 --> 00:01:09.866
Click on the Maximize View button
to close the Tasks pane.

10
00:01:09.866 --> 00:01:10.366
Click on the Maximize View button
to close the Tasks pane.

11
00:01:10.366 --> 00:01:18.099
At a minimum, the One-Way ANOVA task
requires a data set, a dependent variable,
and an explanatory variable.

12
00:01:18.100 --> 00:01:23.900
Click on the icon next to the field
for the data set name.

13
00:01:23.900 --> 00:01:31.100
Select a data set for the analysis by double-
clicking on the name of the library and then 
double-clicking on the name of the data set.

14
00:01:31.100 --> 00:01:37.966
We will use the upsit data set from the libsas 
library.

15
00:01:37.966 --> 00:01:49.099
To specify the dependent variable, click
on the plus sign next to Dependent
variable and then click on smell.

16
00:01:49.100 --> 00:02:01.900
To specify the explanatory variable, click
on the plus sign next to Explanatory
variable and then click on agegroup.

17
00:02:01.900 --> 00:02:10.566
The code window on the right now contains
SAS syntax for running a one-way ANOVA
with the SAS Studio defaults.

18
00:02:10.566 --> 00:02:16.499
Because SAS generates code
for our analysis and then updates
the code as we specify options,

19
00:02:16.500 --> 00:02:22.700
the interactive code generator can
serve as a guide for those who wish
to learn SAS programming.

20
00:02:22.700 --> 00:02:32.766
Though we could run the current syntax,
we will instead examine the OPTIONS tab first.

21
00:02:32.766 --> 00:02:39.766
The OPTIONS tab allows us to specify optional 
tests to be run and to choose plots to be output.

22
00:02:39.766 --> 00:02:48.132
By default, the One-Way ANOVA task
conducts a test for normality, checks for 
homogeneity of variance using Levene’s test,

23
00:02:48.133 --> 00:02:56.633
conducts Welch’s variance-weighted ANOVA,
and performs pairwise comparisons between
all levels of the explanatory variable

24
00:02:56.633 --> 00:03:03.166
using the Tukey method to control
for Type I experimentwise error rate.

25
00:03:03.166 --> 00:03:08.499
These tests can be suppressed, and in some
cases, the testing methods can be changed.

26
00:03:08.500 --> 00:03:08.533
For example, instead of using Levene’s
test for homogeneity of variance, we
might use the Brown & Forsythe test.

27
00:03:08.533 --> 00:03:13.666
For example, instead of using Levene’s
test for homogeneity of variance, we
might use the Brown & Forsythe test.

28
00:03:13.666 --> 00:03:16.299
For example, instead of using Levene’s
test for homogeneity of variance, we
might use the Brown & Forsythe test.

29
00:03:16.300 --> 00:03:22.766
By default, the One-Way ANOVA task
produces a box plot for each value
of the explanatory variable.

30
00:03:22.766 --> 00:03:31.066
Let’s also request diagnostics plots by
clicking the box next to Diagnostics plot.

31
00:03:31.066 --> 00:03:39.532
This is generally a good idea, as the resulting
plots provide a method for graphically
checking for normality and homoscedasticity.

32
00:03:39.533 --> 00:03:52.799
Now that we have specified the options
for our analysis, we run the program
by clicking on the Run button.

33
00:03:52.800 --> 00:03:56.300
The right side of the screen
should now display the results.

34
00:03:56.300 --> 00:04:06.900
In order to better see our results,
we can click on the Code/Results
button in the top left of the screen.

35
00:04:06.900 --> 00:04:15.566
This maximizes the Code/Results window.
To return to the previous view, you could
simply click on Split.

36
00:04:15.566 --> 00:04:25.832
The window now shows the syntax for our 
analysis, rather than the results. To return
to the results, click on the RESULTS tab.

37
00:04:25.833 --> 00:04:30.899
Alternatively, we could click 
on the LOG tab to see the log.

38
00:04:30.900 --> 00:04:35.866
The output begins by listing the
levels of the explanatory variable.

39
00:04:35.866 --> 00:04:41.132
The next table tells us that the
data set contains 180 observations.

40
00:04:41.133 --> 00:04:48.933
Because the data set doesn’t contain
missing data, all 180 observations
were included in the analysis.

41
00:04:48.933 --> 00:04:55.766
Next comes the overall analysis
of variance table, which provides
a test of the overall ANOVA model.

42
00:04:55.766 --> 00:05:02.299
The test is significant, indicating that sense of 
smell significantly differs across the age groups.

43
00:05:02.300 --> 00:05:13.166
Below we have the R-square value, the 
coefficient of variation, the root mean square 
error, and the mean of the dependent variable.

44
00:05:13.166 --> 00:05:17.199
The next two tables provide tests
for each effect in the analysis.

45
00:05:17.200 --> 00:05:28.800
Because this is a one-way ANOVA, however, 
these two tables provide redundant information
from the overall analysis-of-variance table.

46
00:05:28.800 --> 00:05:38.733
Because we checked the Diagnostics
box on the OPTIONS tab, SAS provides
us with a panel of diagnostics plots.

47
00:05:38.733 --> 00:05:45.299
SAS then displays the default box plot.

48
00:05:45.300 --> 00:05:51.266
Levene’s test for homogeneity of variance is
presented next, followed by Welch’s ANOVA.

49
00:05:51.266 --> 00:05:57.966
Because Levene’s test is significant, we should 
favor Welch’s test over the regular ANOVA test.

50
00:05:57.966 --> 00:06:05.999
Like the overall test, Welch’s
test is also significant.

51
00:06:06.000 --> 00:06:14.833
The next section of output contains pairwise
comparisons between all of the age groups.

52
00:06:14.833 --> 00:06:24.033
The final piece of output provides tests
for normality of residuals. All four tests
indicate non-normality of the residuals.

53
00:06:24.033 --> 00:06:31.966
Based on these tests, we may want
to consider re-analyzing the data
using a robust analysis of variance.

54
00:06:31.966 --> 00:06:40.599
This concludes our video on performing
a one-way ANOVA in SAS Studio.