The Utility of Intervening Constructs in Experiments