Variables can be distinguished into experimental and non-experimental variables. The experimental variables are those that are of primary importance because they are the subject of the study. The non-experimental variables are not of primary importance, but can have an impact on the experimental variables.
Types of variables
In research several types of variables are distinguished. An important classifiacation is the level of measururement. A second classification is experimental and non-experimental variables. This is explained in this section.
An independent variable is a variable that the scientist wants to compare. For instance he might wonder if there is a difference between men and women, so the variable is gender (with two values: man or woman). A second example is: Is there a difference between students who attended a course and students who didn’t. Again this is one variable (attending the course) with two values (yes or no). A third example: Is there a difference between countries. This is one variable (country) with a lot of possible values. A fourth example: when growing older people might behave differently. The independent variable now is age.
A dependent variable is a variable that may change due to the independent variable. In our first example (about gender) this may be income, age of marriage, hours spend in the kitchen. In the second example (attending a course) the dependent variable can be something like knowledge, final grade, pass or did not pass. In the third example this might be income, happiness, crime. In the fourth example the dependent variable may be income, health, hours spent on sport.
The relationship between an independent and a dependent variable can be shown graphically as two boxes with an arrow from the independent tot the dependent variable. This kinds of graphs are called the conceptual model.
Most of the time there is a logical relation between the independent and the dependent variable. But the investigated relationship is also arbitrary. It is up to the scientist to assign the variables that are the independent ones and which ones are the dependent. It is even possible to have a serial list of variables on which the variable in the middle is seen as the dependent variable of the first one, and as the independent variable for the last one.An independent or a dependent variable can be measured at a nominal, ordinal, interval or ratio scale. This is another way to classify variables, and this is explained elsewhere in this dictionary. You need to know both classifications because both are determining the type of analysis that is allowed. I have written a paper about it called How to choose the correct statistical test?
If there are experimental variables, there have to be non-experimental variables as well. Experimental variables are of primary interest to the researcher. It is the relationship he wants to investigate. Non-experimental variables are not of primary interest but they may be very important. Non-experimental variables may be very important, that means it has a big influence, potentially important or not important at all. The point is, if you don’t investigate it, nothing can be said about it.
Suppose a research is set up to investigate the relationship between gender and income. It is likely that level of education, years of work experience are of influence. So, one way or another, these variables should be taken into account when you analyse the relationship between gender and income.
In qualitative research it is almost impossible to control the non-experimental influences. Statistical techniques can be used in quantitative research. Now there are different ways to investigate how the non-experimental variables influence the relationship between an independent and a dependent variable. We will not explain it here, because you need to know much more about statistics. If you have sufficient knowledge of statistics, you can find more information on our pages about confounder, mediator, moderator and hierarchical regression.
Related topics to experimental variables:
- Nominal data
- Ordinal data
- Interval data
- Ratio data
- Research design
- Conceptual model
- Hierarchical regression