You might supplement these examples with examples from recent research, such as the finding that the use of artificial bright lights for treating winter depression works well for patients that sleep a lot, have increased appetite, carbohydrate cravings, and who tend to feel worst in the evening, but not as well for patients who have insomnia, anxiety, suicidal tendencies, and who feel worst in the morning. Similarly, a cancer drug that was seen as ineffective several years ago because of a modest 10% success rate, is now held in much higher regard because researchers have found that, for certain types of cancers, the drug is almost 100% effective.
To help students understand interactions, you might want to discuss interactions involving drugs (such examples also emphasize the importance of interactions--it can be a matter of life or death.) If you discuss drug interactions with herbal remedies, you can also point out that using "harmless remedies" can be harmful. For example,
Either Dr. Koop's site (which has an introduction dramatizing the importance of understanding interactions) or the drug digest site (which allows students to choose a drug rather than having to recall drugs from memory) can help students find specific examples of interactions involving prescription drugs. If you want a lighter example of interactions, you can use an example from Mel Brooks' "The Producers." In trying to create a Broadway flop, producers Max Bialystock and Leo Bloom pick the worst script ("Springtime for Hitler'), the worst director, and the worst actors. Unfortunately for the producers, the combination results not in a really bad historical drama but in a hilarious spoof.
Once students understand interactions at the conceptual level, they will still need some practice before they are able to recognize and interpret interactions in data. Specifically, they will have trouble in two areas:
1. interpreting tables of means; andTo give students the necessary practice with tables of means, go over tables 12.5 - 12.11. (If you haven't already, have them go through the advanced factorial design worksheet). If students still have trouble, have them graph the table's results (this may take some time). Then, show them that with interactions, the lines are not parallel. They either cross or, if they were extended, they would cross. Students rapidly grasp this idea. However, make sure that they understand that there is more to interactions than just lines crossing. Reiterate the points made in tables 12.2 - 12.3.
2. confusing main effects with interactions when interpreting ANOVA summary tables.
An excellent article that explains the concept of interactions extremely well is
Astin, A. W. (1993). An empirical typology of college students. Journal of College Student Development, 34
, 36-44.
We often have students get in groups of 2-3 and read discuss the first two pages of this article.
For a more detailed presentation of discussing interactions, see
Schaefer, V. H. (1988). Teaching the concept of interaction and sensitizing students to its
implications. In Ware, M. E. & Brewer, C. L. (Eds.), Handbook for teachingstatistics and research methods. Hillsdale, NJ: Lawrence Erlbaum Associates.