Data Quirk 2: Non-Normality
One assumption in multiple regression is that the residuals are normally distributed (Williams, Grajales, & Kurkiewicz, 2013). There are a variety of ways for the
residuals to fail to meet this assumption, but a common one is for the outcome variable to have a non-normal distribution, such as when it has excessive skew or
kurtosis. For the current example, I made Y's skew equal to negative four and its kurtosis equal to seven. A plot of such a variable is in Figure 6.