suggest an alternative model for the regression equation developed in part (a) to explain as much of the variability in delay as possible