Description:
In this video tutorial, we will briefly explore running the One-Way Analysis-of-Variance procedure in NCSS. Suppose we are comparing the yield from each of three treatments. We wish to determine whether any of the average group yields are different from any of the others.
The data for One-Way Analysis-of-Variance in NCSS can be in either of two formats: one column of responses and one column of groups, or one column for each group. We’ll use the corn yield data with one column per group. After opening the One-Way Analysis-of-Variance procedure from the menu, the Input Type is set to one column per group.
The columns with the yields are then selected.
On the Reports tab we’ll check the assumptions box and the main Analysis-of-Variance report box.
For comparing each group to each other group we’ll check the boxes for the recommended Tukey Kramer multiple comparison test.
To produce the analysis report, the Run button is pressed.
The Normality Tests do not indicate any problems with the Normality assumption. There is also no indication of significantly unequal variances between the groups.
The box plot gives a good visual comparison of the group distributions.
In the Analysis-of-Variance table, a p-value of 0.00175 indicates strong evidence of unequal group means.
The Tukey Kramer Multiple-Comparison Test report shows which groups are significantly different from each other. This report shows that Group C is different from both A and B.
Confidence intervals and p-values for those individual differences are given in the next report.
