ABSTRACT
Research questions often involve more than two variables because researchers are usually interested in the interrelationship among several variables. For example, a researcher may hypothesize that the presence or severity of attention deficit disorder (ADD) is dependent on gender, age, and family history. In this case one would likely be interested in whether the presence or severity of ADD can be explained or predicted by gender, age, or family history, or the interaction of any of these variables. In other words, the relationship among and between these variables is of interest. In other situations, a researcher may want to take into consideration independent variables that may be influencing the relationship among the variables of interest. These variables are referred to as covariates. For example, a researcher may be primarily interested in the relationship between political affiliation and income, but hypothesize that this relationship depends on religious affiliation. In this case religious affiliation would be considered a covariate. In either situation, and with either continuous or categorical variables, to examine the interrelationships among several variables or control for the effects of covariates we typically focus on examining the relationship between two variables of interest while holding the other variable(s) or covariate(s) constant. However, the way in which we accomplish this differs depending on whether the variables are categorical or continuous in nature. In this chapter, we will focus on categorical variables and, specifically, cases that involve three categorical variables. Examples of research questions that can be addressed by the methods presented in this chapter include
Is there an association between gender, political affiliation, and age?r How does age affect the relationship between smoking status and a person’s r ability to breathe normally? Is there an association between education, religious attitudes, and income?r Is there an association between marital status and happiness after controlling r for gender?
