Example: A Test of the Investment Model
The statistical procedures of this chapter are illustrated by analyzing fictitious data from a correlational study based on the investment model (Rusbult, 1980a; 1980b). You will remember that the investment model was used to illustrate the use of a number of other statistical procedures earlier in this book (e.g., in Chapters 8 through 13).
The investment model identifies a number of variables that are believed to predict a person’s level of commitment to a romantic relationship (as well as to other types of relationships). Commitment refers to the individual’s intention to maintain the relationship and remain with a current partner. One version of the investment model asserts that commitment is affected by the following four variables:
| Rewards: | the number of “good things” that the participant associates with the relationship; the positive aspects of the relationship |
| Costs: | the number of “bad things” or hardships associated with the relationship |
| Investment Size: | the amount of time and personal resources that the participant has put into the relationship |
| Alternative Value: | the attractiveness of the participant’s alternatives to the relationship (e.g., the attractiveness of alternative romantic partners) |
The hypothesized relationship among commitment and these four predictor variables is presented as Figure 14.7.
Figure 14.7. One Version of the Investment Model
If these four variables actually have a causal effect on commitment, it is reasonable to assume that they should also be useful for predicting commitment in a simple correlational study. Therefore, the present study is correlational in nature and will use multiple regression procedures to assess the nature of the predictive relationship between these variables. If the model survives this test, you might want to follow it up by performing a new study that is more capable of testing the proposed causal relationships between the variables, perhaps a true experiment. This chapter, however, focuses only on multiple regression procedures.