Estimating agonist efficiency from concentration-response curves