Two examples of 2 X 2 experiments (from Pelham)

 

1.  Sixteen subjects participated in this study.  At the start of the experiment, subjects were randomly assigned to one of two, equally-sized groups:  subjects in one group were put into a positive mood and subjects in the other group were put into a negative mood.  Subjects then were given a list of 20 words to memorize.  One week later, the subjects returned to the laboratory.  Of the eight subjects who had been put into a positive mood the week before, four were randomly selected again and put into a positive mood.  The remaining four were put into a negative mood.  Of the eight subjects who had been put into a negative mood the week before, four were randomly selected and again put into a negative mood. The remaining four were put into a positive mood.  Subjects then were asked to recall the words they had learned the previous week.  (This experiment is called a completely randomized, factorial design with 2 between-subjects factors:  mood at learning and mood at recall.  The dependent variable is number of words recalled.)

 

a.  Draw 2 X 2 grid representing this study.

b.  Based on what you remember from Intro Psych about state-dependent memory, make predictions about what will happen in this study:

      Do you predict a main effect for mood at learning?  If so, what is it?

      Do you predict a main effect for mood at recall?  If so, what is it?

      Do you predict an interaction?  If so, what group(s) will recall the most words?

 

c.  Draw a diagram (line graph) depicting your hypotheses.  If you are predicting an interaction, what sort is it -- spreading or cross-over?

d.  Test your predictions by entering the data below into SPSS and running a 2 X 2 ANOVA. Check your SPSS book for instructions and note that “General Linear Model” is a synonym for ANOVA and “factor” is another term for an independent variable. 

e.  Write a paragraph reporting the results of your ANOVA. Present the appropriate statistical information for the learning main effect, recall main effect and interaction.  Report means for statistically significant effects.

 

Data

Learning Mood    Recall Mood                         Words recalled

positive             positive                         15        

positive             positive                         18

positive             positive                         13

positive             positive                         14

positive             negative                        11

positive             negative                        13

positive             negative                        8

positive             negative                        10

negative            positive                         11

negative            positive                         9

negative            positive                         7

negative            positive                         10

negative            negative                        12

negative            negative                        8

negative            negative                        10

negative            negative                        10

 

2.  In social psychology, there are two competing theories about self-esteem.  One theory posits that people who have low self-esteem prefer negative feedback about themselves. The theory is called self-verification. The reasoning is that a person prefers to maintain a consistent self-view rather than be confused by contradictory information that may prove to be false later. So, people with low self-esteem would prefer negative feedback over positive feedback because it would prevent them from having to change their self-view and from potentially realizing later that that positive feedback was wrong after all.

      A competing theory is called self-enhancement.  This theory posits that everyone is anxious to look and feel as good as possible no matter what. So, this theory would predict that people who have low self-esteem would like positive feedback just as much as people with high self-esteem.

 

Below is the FAKE data from a FAKE study testing these hypotheses.  10 high self-esteem participants and 10 low self-esteem participants were randomly assigned to receive either negative or positive feedback from a confederate (e.g., "You don't seem to be very good at this task; it seems like the kind of thing you're just not cut out for” versus "You did a really great job on this task; you really seem to have what it takes").  The dependent variable was participants' liking for the confederate measured on a 9-point scale (higher scores indicate greater liking).  Thus, this “quasi-experimental” design was a 2 (self-esteem) by 2 (feedback) factorial.  (What makes it “quasi-“ experimental?)  The underlined phrase is a common way to express your design in a method and/or result section.

 

a.  Enter the raw data into SPSS and conduct a 2 X 2 ANOVA.

b.  What do the results indicate?  Are there significant main effects?  Is there a significant interaction? What is the nature of these effects (i.e., what are the means)?

c.  What do the results tell you about self-verification versus self-enhancement theory?  Regardless of what kind of feedback low self-esteem participants apparently preferred relative to high self-esteem participants, can you think of a reason why we cannot call this experiment a test of self-verification theory?

d.  If you had failed to include self-esteem as a factor in this study, and you had simply run a t-test to see if people preferred positive or negative feedback, what would you have found?  Check your answer by conducting this analysis.  Are these results misleading in any way?

 

SELF ESTEEM

FEEDBACK

            LOW                            HIGH

NEGATIVE                 7,8,7,9,9                        3,2,3,4,3

 

POSITIVE                   2,3,4,5,1                        6,9,9,7,9