Huck, S. W. & Sandler, H. W. (1986) Rival >hypotheses: Alternative
interpretations conclusionsof data based
. New York: Harper and
Row.
If you want a less canned, more current set of studies, just glance through the "Crosstalk "section of the latest Psychology Today or research blurbs published in Science News or the USA Today.
If you put the threats on an overhead (you can use table 8.4 [p. 237]), hand out the blurb, and have them students work in teams, students do fairly well. To ensure some success, you may want to alter the write-ups of the first several blurbs to make the flaws more obvious. For example, you may magnify the selection bias by making the groups more different than they were, help students tune in to regression by pointing out that the participants were selected on the basis of extreme scores, draw attention to the mortality rate in the treatment group, etc.
After students find the flaw, you may then ask them for suggestions on how to properly do the study. Their solutions could provide a nice transition into either chapter 9 (the simple experiment) or chapter 13 (quasi-experiments).
Alternatively, you may ask students why people are such suckers for before/after designs (use table 8.3 [p. 236]) and the pretest-posttest demonstration to emphasize the weakness of these designs). Students responses could be used to generate some research studies. For example, a student might see if they had a training program that could make people less convinced by pre/post designs. Testing the effectiveness of such a program would not only give students a viable research project, but would also help them remember the problems with before-after designs.