The Games-Howell method compares the means for each pair of factor levels using a family error rate (often called familywise error rate) to control the rate of type I error. The family error rate is the probability of making one or more type I errors for the entire set of comparisons. The Games-Howell method adjusts the individual confidence level, based on the family error rate you choose.

This method is only available if you do not assume that all populations have equal variances.

Use the confidence intervals to determine likely ranges for the differences and to assess the practical significance of the differences.

·    If an interval does not contain zero, there is a statistically significant difference between the corresponding means.

·    If the interval does contain zero, the difference between the means is not statistically significant.

To display the values of the confidence limits in the Session window, check Tests in Stat > ANOVA > One-Way > Comparisons.

For the paint hardness data, the confidence intervals display the likely ranges for all the mean differences:

·    The confidence interval for the difference between the means of Blend 2 and Blend 4 extends from 1.31312 to 17.6869. This range does not include zero, which indicates that the difference between these means is significant.

·    The confidence intervals for the remaining pairs of means all include zero, which indicates that the differences are not significant.