Chapter 3 introduces asymmetry properties of the linear regression model that emerge from distributional properties of non-Gaussian variables. Foundations, significance procedures, and decision guidelines of third and fourth moment-based DDA measures are discussed for two scenarios. In the first scenario, we assume that the error term of the “true” model follows a Gaussian distribution. In the second scenario, we relax this distributional assumption and discuss direction of dependence based on distributional characteristics of observed variables when the “true” error term deviates from normality. Further, power characteristics of the DDA measures are presented based on a Monte-Carlo simulation experiment, and synthetic as well as real-world data examples are given to illustrate DDA model selection based on observed variable distributions.