Explaining Variation in Instructional Time: An Application of Quantile Regression