Say you have data on annual income for 1,000,000 residents of the U.S. You want to estimate a 50% prediction interval, that is an interval for the middle 50% of incomes. This could provide you some sense of what the range of incomes are for middle class earners. The data is not normal and thus the formulas provided in lecture 2 will not work. Without transforming or manipulating the data in any way, how could you estimate a 50% prediction interval?