Managers often assume a strictly linear relationship between cost and the level of activity. How can this practice be defended in light of the fact that many costs are curvilinear?