The two active methods of control (holding constant and matching) require some extra effort or extra measurement and, therefore, are typically used with only one or two specific variables identified as real threats for confounding. In addition, holding a variable constant has the disadvantage of limiting generalization (external validity).