Variable

Variable Attributes

All variables must include at least two distinct values that differ across research subjects. These values, or categories, are the characteristics that describe the item of interest. For example, a fundamental component of the basic experimental design is the treatment variable. This measure consists of at least two possible categories: “treatment” and “no treatment” (or “control”). A more complex experimental design might also include a “placebo” group as a third value. Quite commonly, variables can include substantially larger numbers of values ranging from modestly more complex Likert scaling, which typically uses a five-category coding from “strongly disagree” to “strongly agree,” to extremely detailed measures such as IQ or income. These more detailed measures are frequently encountered when experimental design is not feasible and when researchers use some form of survey research.

Regardless of how detailed and complex the attributes of any variable becomes, all variables are subject to two requirements. First, the values for any variable must be mutually exclusive from one another. In other words, each subject can have only a single value on each variable in the study. For example, it would be inappropriate for the values of a variable to be male, female, and Latino because it is possible (and, in this case, quite likely) that many individuals might be both male and Latino or female and Latino. On the other hand, a variable that includes the more logical grouping of only male and female categories would satisfy the mutual exclusivity requirement because one cannot simultaneously be a male and a female.

Second, the attributes of any variable must be exhaustive. This requirement does not mean that every [Page 1787]possible value of a variable must be included in the measure. Rather, this requirement is satisfied if all research participants can be assigned, or self-selected, into a value provided. For example, a variable that measures a subject’s race might include values such as “white,” “black,” “Asian,” and “Native American/American Indian.” The categories in this case would not be exhaustive because some individuals, such as a person of Australian Aboriginal descent or a multiracial person, would not fit into any single category. A simple fix in this particular case would be to add a “multiracial” category as well as an “other” category. The inclusion of residual categories like “other” in the preceding example is commonplace when such values represent extremely small portions of the population being studied or when the actual values of the residual groups are not directly relevant to the research question at hand.

...