|
Subject to concerns about not overdoing the number of IVs in your experiment, when considering controlling the important variables in an experiment, think about whether it might be possible and useful to manipulate the variable instead. For instance, in the cheese and nightmare experiment in Chapter 9 of the book, we decided to exclude people who drank alcohol from the study (Section 9.1.5). We did this because we were concerned that alcohol might influence the number of nightmares that people experienced and we wanted to eliminate this source of extraneous variation from our study. An alternative would have been to have randomly allocated participants either to not drink alcohol or drink some alcohol (say 1 or 2 standard sized glasses) prior to sleep in addition to manipulating whether they either ate or did not eat cheese. That way we could test the effect of the cheese manipulation at the levels of the alcohol IV and learn more about its effects. We would know whether the effects of cheese were the same regardless of whether people drank (small amounts of) alcohol or whether these IVs interacted in some way. (For more on interactions see Section 13.5 of the book and C2 of this Web site.) Of course, it may not be feasible or ethical to manipulate the variables that you feel you otherwise need to hold constant. However, there may be occasions when you can do so simply and without overloading your experiment with IVs. (See Section 13.6 of the book for a discussion about why you should not overload your experiment with IVs.) |
