A simple random sample of 100 of a certain popular model car in 2003 found that 20 had a certain minor defect in the brakes. A simple random sample of 400 of this model car in 2004 found that 50 had the minor defect in the brakes. Let p1 and p2 be the proportion of all cars of this model in 2003 and 2004, respectively, that actually have the defect. What is a 90% confidence interval for p1 - p2?