In this article, we discuss and propose methods that may be of use to determine direction of dependence in non-normally distributed variables. First, it is shown that standard regression analysis is unable to distinguish between explanatory and response variables. Then, skewness and kurtosis are discussed as tools to assess deviation from normality. Deviation from normality can be used to assess direction of dependence. This proposition is based on the fact that the response variable will always have less skew than the independent variable (Dodge & Rousson, 2000). It has been shown that the cube of the Pearson correlation coefficient can be calculated as the ratio of the skewness measures of the correlated variables. Because correlations cannot exceed the interval (-1.0; +1.0), directional dependence of the two correlated variables can be determined by the ratio that results in a correlation that stays within this interval. It is also proposed that other measures of deviation from normality can be used to determine directional dependence; for example, kurtosis. Recommendations are given for making decisions concerning directional dependence. Empirical data examples from developmental research on violence against women and on attention deficit hyperactivity disorder (ADHD) illustrate the use of the methodology. Cross-sectional and temporal directional dependence are discussed, and the effects of onset of a causal agent and termination of a causal agent are illustrated. (Contains 5 tables and 3 figures.)