Abstract
We consider functions which are the sum of the k largest order statistics in a sample of size n from a continuous distribution F, minus nh, where h is a specified constant. We prove that such functions are concave in n. If F is an exponential distribution, then for a fixed k we obtain that value of n which maximizes the expected value of the function defined above. For F IFR we obtain an upper bound on n and also an upper bound on the maximum of the expected value of the function.