I. Use a factorial design to improve generalizability by
A. Adding a replication factor (stimulus set) to see if the results generalize to other stimulus materials
B. Using a situational variable as a second factor to see if the results generalize to different situations
C. Using participant type as a second factor
1. Pro: Useful for seeing whether the results really do generalize to a particular group
2. Con: One can't make causal statements about participant variables because participant variables cannot be randomly assigned
II. Use a factorial design to replicate a published study AND show that the effect does not hold in certain circumstances: Moderating factor designs
III. Use a factorial design to improve power: The blocked design to siphon off error variance. Logic:
IV. The factorial design as a way to study certain variables.
A. An interaction may represent similarity: (e.g., participant-experimenter similarity, similarity between learning-recall context)
Place Information Learned Place Tested for Information Basement Downstairs Basement 10 5 Downstairs 5 10 Words Recalled
B. An interaction may give you a more accurate picture of what the treatment variable is really manipulating. For example, suppose you did a simple experiment and found that men rate a woman who wear a certain scent more favorably. You might conclude that the scent's effect is to increase general attraction. However, if you did a 2 X 2 factorial and found an interaction indicating that the scent increases ratings of women, but not of men, then you might conclude that the scent intensifies sexual attraction. Similarly, if you did a simple experiment and found that positive ions made happy people feel happier, you might think that positive ions elevate mood. However, if you did a 2 X 2 and found an interaction such that positive ions made happy people feel happier, but tense people feel more tense, you would conclude that positive ions intensify existing mood rather than elevating existing mood.
Back to Chapter 12 Main Menu