A control variable is a variable that might influence the relationship between the experimental variables. Because this is not of primary interest for the topic of the research it should be held constant or neutralized.
When is a control variable needed?
It can be interesting to investigate the relationship between job satisfaction and the intensity of looking for a new job. The hypothesis is that people who are more satisfied are not looking for a new job, while people who are more dissatisfied more often look for a new job. Sounds reasonable, does not it?
People who are older, closer to their pension, are probably less interested because they will have to move to a new firm. People who have a working partner are not willing to move, and that will reduce the possibilities and probability the intensity for looking for a new job. If the person has children attending a primary or secondary school, they will probably protest if they have to move to another city. This will also reduce the intensity in looking for a new job. All these aspects should be taken into account when a study is done about the relation between satisfaction with the job and the intensity of looking for a new job.
How is a control variable used in statistics
In this example job satisfaction and intensity of searching for a job are treated (operationalised) as variables on interval or ratio data. The statistical test to investigate this relationship is a correlation. To take the aspects of age and family aspects into account, a regression has to be performed. To treat them as control variables these variables should be put in first in the equation and then the predictive variables. Read more about this on our pages regression and hierarchical regression.