Consumer choice occurs over multiple products and services, each comprising multiple risks. In this paper, we present a new market research technique to measure consumers' preferences over large spaces of risks. We first describe the method, present its psychological and analytical motivation, and then report the results of empirical tests of reliability and validity, both within testing sessions and across the span of one year. The method is used to estimate the coefficient of relative risk aversion and the loss aversion parameter for a sample of adults saving for retirement. The new technique passes tests of reliability and validation and captures individual differences based on age and income. It also identifies two sub-populations, one best fit by a more classical model of risk preference, and the other by a behavioral model which incorporates loss aversion.