Adverse Childhood Experiences: Navigating the Challenges of Statistical Bias & Causal Inference

• In Event: Expanding Statistical and Methodological Approaches in DLC Work

Abstract

Developmental criminologists have grown increasingly focused on the topic of adverse childhood experiences (ACEs) in recent decades, and with good reason. Mistreatment, neglect, and abuse early in the life-course is correctly understood to be a risk-factor for a range of internalizing and externalizing problems as children age. Scholars who work on this topic are also well acquainted with the challenge that it represents from a methodological standpoint. The observational data required to study the effects of ACEs has become increasingly available, so the problem, in many respects, is not one of availability. Instead, it is the ever-present difficulty one faces when left to search for causal effects in non-experimental designs. There are two central concerns on this front: confounders and colliders. Both can create highly misleading results, but only one (confounding) is somewhat widely appreciated. The discussion here centers on the usefulness of family designs for the study of ACEs and explores various options for addressing both confounding and colliding in data. Having surveyed these subjects, the discussion concludes by reviewing the existing work on ACEs conducted so far with family-based methods and explores some implications for not only future empirical work, but for theory building as well.

Author

• Brian Boutwell, University of Mississippi