The variable the experimenter manipulates.
In experimental research, psychologists create two or more groups that are as similar as possible except for a single change that the psychologist makes from one group to the next. That single element that varies across groups is called the independent variable. In more complex research, the experimenter may include more than one independent variable.
In one experiment dealing with eyewitness testimony and jury decisions, researchers exposed the eyewitnesses to staged crimes and then had them "testify" what they observed. One group of participants saw the staged crime under good lighting conditions; a second group had a less favorable viewing condition, and the third group had only a poor view of the scene. The independent variable was the viewing condition which had three levels, or different variations: good, moderate, and poor visibility. The researchers investigated whether the "jurors" accepted the testimony as believable and the degree of confidence of the eyewitnesses in their own testimony. The degree to which the jurors accepted the testimony and the stated degree of confidence by the witnesses themselves were dependent variables. The results revealed that the jurors were more likely to believe witnesses who had seen the crime in the best lighting.
The researchers concluded that the independent variable (e.g., the amount of light available for viewing the crime) had affected one dependent variable (e.g., the jurors' acceptance of the testimony). At the same time, the independent variable did not affect the confidence of the eyewitnesses concerning their own testimony.
Lindsay, R. C.; G. L. Wells; and C. M. Rumple. "Can People Detect Eyewitness Identification Accuracy Within and Across Situations?" Journal of Applied Psychology 67 (1981): 79-89.