Main effects and Interactions

Answers at the end of the page.

1. Look at the table below.

A1 A2
B1 20 20
B2 30 30

a. Is there a main effect for A? Why or why not?

b. Is there a main effect for B? Why or why not?

c. Is there an interaction? Why or why not?

2. Consider the table below.

A1 A2
B1 20 30
B2 30 40

a. Is there a main effect for A? Why or why not?

b. Is there a main effect for B? Why or why not?

c. Is there an interaction? Why or why not?

3. Examine the table below.

A1 A2
B1 20 30
B2 30 20

a. Is there a main effect for A? Why or why not?

b. Is there a main effect for B? Why or why not?

c. Is there an interaction? Why or why not?

4. Examine the table below where the d.v. is percent correcton a test of math or verbal ability

Math Verbal
No Training 100 20
Training 100 80

What can we say about the effects of training for math relative to verbal ability?

Why might we distrust the observed interaction?

5. Consider the table below.

Children Adult
Reward 20 30
Punishment 10 80

a. Is there a main effect for Age? Why or why not?

b. Is there a main effect for Reward/Punishment? Why or why not?

c. Is there an interaction between the two? Why or why not?

 

Answers

1. a. No b. Yes c. No.

B results in an increase as you change its level from 1 to 2 . Changing from level 1 to 2 of A, however, results in no change in the d.v. The size of the change in the dv for B does not depend on the level of A (no interaction - also, plotting shows the lines are parallel).

2. a. Yes b. Yes c. No.

Both A and B result in an increase in the dv as one changes from level 1 to level 2 for each variable (2 main effects). The size of this increase(10) for both IVs does not depend on the level of the other IV (plotting also shows the lines to be parallel).

3. a. No b. No c. Yes.

The average effect of each variable is zero. Same is true for B. This is an interaction without main effects.

4. Training improves verbal but not math test scores. Distrust these results because of a possible ceiling effect - Math ability was already at its maximum before training was introduced

5. a. Yes b. Yes c. Yes

Adults always have higher dv scores than children (a main effect for Age). Reward results in higher scores for children (vs. punishment) on the average. The size and direction of the effect of reward, however, depends on the level of the other variable (age) indicating an interaction is present (plotting also reveals that the lines are not parallel).