In general, scientific research tests a hypothesis by manipulating a single variable and examining its effect on another variable. The manipulated (or independent) variable is typically altered in a number of different ways, and the responding (or dependent) variable is what is examined or measured over the course of the experiment. In order to yield accurate results and draw conclusions from them, all other aspects of the experiment must be tightly controlled and kept as close to the same as possible for every trial. This ensures that no factors other than the independent variable are influencing the dependent variable.
This paper (link below) examines the effect of subliminal self-help audiotapes on the weight of females. There were three groups in the study. The first group listened to the subliminal self-help tapes, the second group listened to a placebo (tapes other than self-help tapes), and the third group listened to no tapes. Because the researchers changed the presence or absence of different types of audiotapes, this is the independent variable.
The researchers also measured the weight of each subject over the course of the experiment; because this is what they were keeping track of, this is the dependent variable. It is tempting to say that weight loss is the dependent variable, but this is inaccurate, because the subjects were not guaranteed to lose weight. It’s more accurate to say that the weight of each subject is the dependent variable, because this is what the researchers were directly measuring.