Week14 & 15
Multivariate and within-subjects designs.
So far we have been talking about simple designs that assign subjects to one group or the other (i.e., control or experimental) . These are between-subjects designs. We will discuss different conditions when it is more informative for subjects to be in more than one condition, or to be measured more than once: "within subject" designs.
Understand the key features of these designs:
- Own control
- Each participant is in both the control and experimental groups.
- Makes the contrast between conditions very clear & strong.
- Order of groups must be counterbalanced.
- Reversal designs
- Test the hypothesis that a behavior is controlled by a specific temporary condition.
- Test it by imposing the condition (e.g., stress) on people, then withdrawing it, then imposing it again.
- Design: A - B - A / Impose - withdraw - Impose condition.
- Repeated measures and randomized block designs
- Multiple treatment conditions: each participant gets each treatment, similar to an "own control" design..
- Longitudinal / time sampling: each participant assessed over multiple time periods.
- Randomized block designs; Repeated measure combined with between-groups variable, e.g., an experimental group and a control group, each measured over time.
Your Chapter on Within-Subjects and Matched Subjects Approaches (Chapter 10 in Ray) & discussion of interpreting studies with multiple measured and unmeasured variables, and readings on interpreting “relative risk” statistics, and interpreting data about risks + benefits of mammography.
Discussion group Assignment
Develop a multivariate design
This week you will turn your study into a factorial or randomized block design. Take a look at the lecture notes for examples of multivariate studies. Model your study on one of these designs if that helps, but think of something original.
- Take your paper topic and reframe it as having two (or more) Independent Variables. You already specified one IV for your paper. Now think of another IV; you could add
- A true Independent variable – a second manipulation, in a factorial design;
- A blocking variable, such as a status variable (age group, gender, ethnicity...); or
- A within-subject variable, where you follow participants over time or have participants in > 1 experimental condition (e.g., each participant gets both the experimental and control condition).
- Write the main effect hypotheses. There will be one for each independent variable.
- You already have written a hypothesis for your study IV.
- Now write main effect hypotheses for the variable(s) you are adding..
- Write the interaction hypothesis.
- Does one variable have an effect on the Dependent Variable only at one level of the other IV?
- Does one variable have opposite effects on the DV at each level of the second IV?
- Sketch out a graph that illustrates data that would support the interaction hypothesis.
- What do you know by testing the two (or more) variables together that you might not have detected by testing one at a time?