745 / 762
Chapter 16: One-Way Analysis of Variance
745
In the mathematical model, it is assumed that the errors
are independent of each other.
are normally distributed.
have a variance which is equal to a constant variance for the
potential populations of responses for the different treatment
levels.
Thus we can use the errors in the model to help validate the ANOVA
assumptions.
Note that “constant variance” here means that the variances for the different
potential sampling populations are equal to each other.
Example 16-12:
For the percentage brick heat loss data for the non-toxic
chemical mixes, validate the assumptions for the model.
Solution:
When validating the one-way ANOVA assumptions we can
make use of the computer and readily available workbooks and statistical
software. We will check these assumptions by using the error values for the
model
.
Validating the Normality Assumption for the Model Errors
First we need to compute the errors or residuals for the data in the example.
We can use the
Auto Correlation Function
workbook to compute the