Nested versus Non-nested ANOVAs

Nested ANOVAs are for cases where a fully crossed ANOVA cannot occur. The most simple example is when studying something where random assignment is not possible or a participant cannot participate in all conditions. For example, an ANOVA looking at dog behavior by trainer by training school. It does not make sense to test this data using a crossed design because each dog can only have one trainer and each trainer can only train at the place they're employed to train. Thus in this case, a nested ANOVA would be used with dog behavior nested in trainer nested in school.

By contrast, a non-nested design might have each trainer teach the dog a new command and use a latency measure to see which trainer trained the fastest. This would eliminate nesting as all dogs would use all trainers.

In a nested ANOVA, the null hypothesis has levels. At each level, the null hypothesis is that there is no difference in the group of subgroup.

For more information on nested and nested verus non-nested ANOVA designs, see the following

Back to nested ANOVA