In physics experiments, a great deal of effort is spent calibrating instruments. These include instruments that precipitate some event, and instruments that measure the effects of those events. Design research in the learning sciences often focuses on precipitating learning events, but it does not pay equal attention to designing effective measures. We present the results of a study that compared two types of instruction on students working alone or in pairs. We show how one measure, common to most studies of learning, failed to detect any effects. Then we show how a second measure, called a Preparation for Future Learning measure, detected important differences. Specifically, pairs working to invent solutions to problems in statistics were more prepared to learn about new, related types of statistics than pairs who were shown how to solve the original problems, as well as individuals who invented or were shown how to solve the original problems.