Why do economists use the midpoint method to calculate the price elasticity of demand between demand/price combination (q1, p1) and combination (q2, p2)?