Understanding and Applying Research Design

10

REGRESSION¹

Researchers attempt to understand social problems using specialized statistical procedures. When we explored the use of contingency tables, for example, we saw that we could use chi square to help us determine whether there were meaningful patterns among categorical data. We also saw that contingency tables could be elaborated by adding more independent variables. Social problems are complex and therefore require methods that allow us to recognize and understand this complexity. Adding more variables to the contingency table is one way to see how multiple influences on a dependent variable measure can provide a deeper understanding of the phenomena.

Multiple linear regression (MLR) procedures follow the same logic. This statistical procedure assesses the influence on an outcome variable of more than one predictor variable. When we use correlation procedures, we can understand how one variable influences or is related to another variable. For example, job stress is a pervasive problem in some companies. Not only does it have an impact on productivity and worker satisfaction, it may also influence what happens at home. As Melvin Kohn (1977) demonstrated, the nature of work influences how people respond to one another within their families.

In the real-world context of research, understanding a social phenomenon is rarely achieved only by examining one potential influence. Thus, for example, if we wanted to understand worker stress in more depth, we would probably be ...