ABSTRACT
Multiple regression is a frequently used statistical method for analyzing data when there are several independent variables and one dependent variable. Although it can be used in place of analysis of variance, it is most commonly used in the associational approach. For example, Logan and King (2002) were interested in parents' ability to identify signs of depression in their adolescent children. They hypothesized that prior use of mental health services, impact of the adolescent's emotional/behavioral problems on the family, presence of substance abuse, adolescent-parent communication, and parental depression would be related to a parent's ability to identify depression in the adolescent. In this example, the independent variables, which are referred to in multiple regression as predictor variables, are service use, impact on family, presence of substance abuse, adolescent-parent communication, and parental depression. The dependent variable, which in multiple regression is called the criterion or outcome variable, is parental ability to identify depression. Multiple regression was appropriate for this analysis because the variables are approximately normally distributed (some predictor variables could be dichotomous), and the research question asked how the many independent variables combined to predict the dependent variable.
