Frequently, tests that yield abnormal results are repeated for confirmation. What is the probability that for a usual person a test will be at least 1.5 times as high as the upper limit of normal on two divided occasions?