You are on page 1of 21

12/6/2013

Three-Way Factorial ANOVA (2 2 2 Factorial)

Example
A researcher is interested in determining the effects of the two types of learning strategies (A and B) on the memorization of easy-vs.difficult list as well as in determining the effects of high vs. low intensity shock on learning. The researcher records the total number of errors made by each subject.

12/6/2013

The three-way factorial combination of the three independent variables yields the following eight experimental groups, with six subjects per group

Enter data using the following:

12/6/2013

12/6/2013

12/6/2013

12/6/2013

12/6/2013

6. To plot the three-way STRATEGY*LIST*SHOCK interaction, some data transformation must be carried out. Suppose the researcher decides to plot the variables of STRATEGY and LIST (i.e., STRATEGY*LIST interaction) against the variable of SHOCK. The first step is to create a new grouping variable called GROUP that contains the four levels generated by the STRATEGY*LIST interaction (Strategy AEasy List, Strategy AHard List, Strategy BEasy List, and Strategy BHard List).

12/6/2013

12/6/2013

SPSS OUTPUT

12/6/2013

10

12/6/2013

11

12/6/2013

Results and Interpretation

12

12/6/2013

Main Effects
The main effect of STRATEGY is significant, F(1,40) = 95.16, p < .001. From the estimated marginal means, subjects made significantly more errors under strategy B (M = 25.71) than under strategy A (M = 14.00) (collapsing across the LIST and SHOCK factors). The main effect of LIST is significant, F(1,40) = 121.10, p < .001. Subjects made significantly more errors in the hard list condition (M = 26.45) than in the easy list condition (M = 13.25) (collapsing across the STRATEGY and SHOCK factors). The main effect for SHOCK is not significant, F(1,40) = 1.16, p > .05. The difference in the number of errors made under the low-shock condition (M = 19.21) is not significantly different from the number of errors made under the highshock condition (M = 20.50) (collapsing across the LIST and STRATEGY factors).

Two-Way Interactions

13

12/6/2013

Strategy*List Interaction
The STRATEGY*LIST interaction is significant, F(1,40) = 14.32, p < .01. The significant interaction effect indicates that the effect of learning strategy on the number of errors made is dependent on the difficulty of the list learned. Although the number of errors made increased from Strategy A to Strategy B when learning either hard or easy list, the increase is more pronounced when learning the hard list than the easy list.

14

12/6/2013

Post Hoc Comparisons

15

12/6/2013

Results and Interpretation


Post Hoc Comparisons

Results from the post hoc comparisons indicate that the significant STRATEGY*LIST interaction is due primarily to subjects making significantly less errors in the Strategy AEasy List condition than in the other three experimental conditions (AHard, BEasy, and BHard), and to subjects making significantly more errors in the Strategy BHard List condition than in the other three experimental conditions (BEasy, AEasy, and A Hard).

16

12/6/2013

STRATEGY*SHOCK Interaction

STRATEGY*SHOCK Interaction
The STRATEGY*SHOCK interaction is not significant, F(1,40) = 3.13, p > .05. As the interaction is not significant, the result can be interpreted in terms of the significant main effect for STRATEGY. That is, the effect of STRATEGY on the number of errors made is not dependent on the levels of SHOCK, such that regardless of SHOCK level, subjects made significantly more errors under Strategy B (M = 25.71) than under Strategy A (M = 14.00).

17

12/6/2013

LIST*SHOCK Interaction

18

12/6/2013

LIST*SHOCK Interaction
The LIST*SHOCK interaction is not significant, F(1,40) = 1.47, p > .05. The effect of LIST on the number of errors made is not dependent on the levels of SHOCK, such that regardless of SHOCK level, subjects made more errors on the hard list (M = 26.45) than on the easy list (M = 13.25).

19

12/6/2013

Three-Way Interaction

STRATEGY*LIST*SHOCK Interaction
The three-way interaction between STRATEGY, LIST, and SHOCK is not significant, F(1,40) = 0.27, p > .05. As the three-way interaction is not significant, it is legitimate to interpret the significant main effects of LIST and STRATEGY. For example, the results indicate that more errors were made on the hard list (M = 26.45) than on the easy list (M = 13.25), and under Strategy B (M = 25.71) than under Strategy A (M = 14.00), regardless of shock level.

20

12/6/2013

21

You might also like