Dependent and Independent Variables

One of the fundamental distinctions in medical and epidemiological research is that between independent and dependent variables. In the simplest sense, the dependent variable is the result or outcome being studied and the independent variables are factors that are assumed to exert an influence on it. These basic concepts are very simple but can become confusing in practice, particularly since different researchers use different terms for the same concept. In addition, because some researchers believe that the terms independent and dependent variable imply a causal relationship, they prefer to use one set of terms for experimental research (where it is possible to hypothesize causality) and another for observational research (where due to the number of uncontrolled influences on the outcome, some researchers prefer to speak of correlations or other observed relationships without labeling them as causal).

In experimental research, if a variable under investigation is to be described in terms of other variables, then this variable is called a dependent variable (or response variable or outcome variable). A variable that is believed to influence a dependent variable is called an independent variable (or explanatory variable, or causal variable, or input variable). When data are displayed graphically, the dependent variable is generally represented on the y-axis and the independent variable(s) on the x-axis.

In a standard regression equation of the form

the Y-variable is the dependent variable and the X-variables are the independent variables. Studies often include numerous independent variables and may also include multiple dependent variables.

To take an experimental example, a researcher randomly assigns subjects to receive one of five different oral contraceptives to study their effect on highdensity lipocholesterol (HDLC), a substance found in blood serum. It is believed that high levels of this substance help delay the onset of certain heart diseases. In this case, the five oral contraceptives are independent variables, and the HDLC is the dependent variable. Other independent variables might be included in the study, including the participants’ age, weight, and previous medical history.

The very language of independent and dependent variables implies causality (the value of the dependent variables is assumed to depend in some way on the value of the independent variables), and researchers do sometimes refer to independent variables as causes of change in the value of the dependent variable and [Page 261]changes in the dependent variable as effects caused by the independent variables. However, particularly in observational studies, causality cannot be assumed simply because a relationship exists between two variables, and for this reason, some researchers prefer to use the terms predictor variable for independent variable and criterion variable for dependent or outcome variable in nonexperimental research. The reason is that one variable can predict the value of another in the sense that there is an observable relationship or association between the two, without implying the causal relationship that exists between them. For instance, in the United States, race is associated with poorer outcomes on a number of health measurements, but that fact is not taken as proof that race is the cause of the poorer outcomes.

...