Attribute Variable / Passive Variable: Definition, Examples

Types of Variables >

What is an Attribute Variable?

The term “attribute variable” can refer to either an input variable used in statistical software or a manipulated variable used in design of experiments.

1. Usage in Software

Some statistical software packages (like Statistica and Minitab) use “attribute variable” as an alternate name for categorical variables.

2. Usage in Experimental Design

An attribute variable (sometimes called a passive variable) is a type of variable that is not manipulated in experiments. An attribute variable could be a variable that is a fixed attribute like sex, race, or gender; These variables cannot be changed or manipulated by the researcher as they are an inherent part of a person or object. Or, attribute variables can also be variables that aren’t manipulated by the researcher for a particular experiment (but theoretically, they could be for a different experiment).

The opposite of an attribute variable is an active variable, which is manipulated in an experiment.

Although the definition of an attribute variable includes that the variable is “manipulated”, these variables are not the same thing as independent variables. In fact, the nuances of the definition are very subtle, which can lead to confusion. A few examples should make the definition clear:

Examples of Independent and Dependent Attribute Variables

Attribute variables and active variables can be independent variables or dependent variables.

Example 1: Let’s say you are investigating whether there is a link between taking Vitamin C and preventing illness. Vitamin C is the independent variable, and you run an experiment with an experimental group (who are given Vitamin C) and a control group (who are given a placebo). You are manipulating the Vitamin C, so it is an active independent variable.

Example 2: A retrospective study investigates the Vitamin C/illness link. Patients at an urgent care clinic are divided into two groups — those who took Vitamin C and those who did not. Vitamin C is still the independent variable, but as it’s not being manipulated it’s an attribute independent variable.

Example 3: A researcher wishes to measure how well two teaching styles affect students’ performance on tests. How well the learners perform is not only dependent on teaching style, but on a whole array of attribute variables like socioeconomic status, learning disability, or amount of sleep the students got the night before. These variables, which affect the outcome in some way and are pre-existing conditions not manipulated by the researcher, are attribute dependent variables.

Example 4: In the above teaching style example, researchers “prescribe” an amount of sleep: 5 hours, 7 hours, or 9 hours. As this dependent variable is being manipulated, it’s an active dependent variable.

When Passive Variables become Active

A variable that is passive in one experiment can easily become active in another.

For instance, consider a research project on the use of rice cookers in Asian American homes. In order to find out how rice cooker use relates to healthy eating, researchers surveyed a number of families.  They separated the families into groups based on the number of times per week those families reported using the rice cooker. Survey questions were used to bring to light health differences between these groups. In this experiment, rice cooker usage would be a passive variable. It happens to be the variable of interest to the researchers, but they aren’t manipulating it or imposing values on it in any way.

However, these same researchers might find it useful to design another experiment. Now they take a number of Asian American families and ask one group to use their rice cookers twice daily.  The researchers ask another group to use their rice cookers twice a week and a third group to not use them at all. After some time, the same health survey is given to all participants.

Now rice cooker usage is a manipulated variable, and is active rather than passive.

Note that the data collected in these two experiments  will almost certainly be different. Experiment 1 investigates the health of people in relation to the choices they make or habits they have regarding rice cooker use. But Experiment 2 will highlight any differences actual rice cooker usage has on health, and eliminates self-selection bias.


Basics of Research for the Health Professions: Independent Variables
Game Theory and Economic Behavior

Comments? Need to post a correction? Please Contact Us.