The second model compared the control section, spring 2012, with the second flipped ALC treatment section, spring 2013. For this model, casewise diagnostics were again generated and examined to locate outliers in the data set, and again only a single outlier was found, which was retained in the data set. VIF statistics indicated very little multicollinearity, and the DurbineWatson statistic of 1.378 showed only a moderate amount of positive autocollinearity in the data.