I heard an economist say that in an experiment he "controlled for income level."?

Question

How does controlling for something affect the design of an experiment?

Answer 1

Income was the controlled, dependent, variable. Other things in the study varied, such as education, for example.

Answer 2

Controlling for income would mean that the experimenter(s) had an item that asked participants about their income level. They would have then ran statistical analysis to figure out if income was correlated with what ever outcome they were actually interesting in. If the correlation was insignificant (determined by alpha, which is determined by the experimenter) then income would not be considered a variable with much impact on whatever the economist(s) were studying.

Basically, it is a check to see if a control/test variable (income) is related to whatever outcome is being tested(whatever the study was investigating). I.e. will an increase in income level be related to an increase or decrease in the outcome? Sometimes it is used for testing for a spurious realtionship. Statistical books or research design books will go into greater detail. E.g. "The Practice of Social Research" by Earl Babbie.