Challenges in Establishing Causal Validity of Studies on Influencing Factors in Education

One of the popular approaches to educational research is to investigate multiple influencing factors on an outcome of interest. Researchers ask, “Why do students drop out of school?” or “Why did the math score decrease?,” and try to find potential factors that lead to such consequences. The findings are then used to develop an appropriate policy to intervene to facilitate a desirable outcome or to inhibit an undesirable one. This approach belongs to a type of question referred to as “causes of effects,” which is contrasted to “effects of causes.” Interestingly, many causal inference researchers believe that the former is much more difficult to answer than the latter. However, the theoretical basis for such a belief has not been well discussed in the Korean educational research field despite the popularity of the approach. Using causal graphs, this article explains why the causal interpretation of the findings from many studies on influencing factors cannot be established with certainty. In fact, multiple regression coefficients from a single equation could have different interpretations, such as one as a direct effect and the other as a confounded noncausal association. It is emphasized that with the lack of knowledge of true data-generating process, it is almost impossible to correctly interpret statistical findings in terms of causal validity. The article concludes with a discussion of the importance of causal inference for educational research.