To draw cause and effect conclusions, one needs to conduct a formal experiment, sometimes called a control experiment. The independent variable in this type of experiment is the only thing that is allowed to changed so the experimenter is able to conclude that the it is the independent variable which affected the dependent variable. In other words the independent variable effects the dependent variable, and it is the only thing that can effect the dependent variable. This is not true in correlational experiments. Remember the oft repeated phrase that correlation does not mean causation. In other words, a lot of people carry umbrellas on a rainy day (there is a correlation between rainy days and people carrying umbrellas) but the umbrella carrying people did not cause the rain.