Measuring risk aversion with lists: a new bias