The Impact of Paternal Incarceration on Boys’ Delinquency: A New Method for Adjusting for Model-Driven Bias

• In Event: New Frontiers in the Research on the Consequences of Paternal Incarceration

Abstract

Paternal incarceration has a range of deleterious effects on children, especially on boys’ behavioral problems, and the damaging impact of paternal incarceration on boys’ delinquency is among the most well-established findings in studies of the intergenerational consequences of incarceration. Measures of sons’ delinquency are often dichotomous, indicating whether a son exhibits delinquent behavior, and scholars of criminology often apply nonlinear probability models to analyze such outcomes. But in so doing, we involuntarily make our estimates vulnerable to model-driven bias, because these models are sensitive to their own fundamental assumptions. In this paper, we introduce to scholars of criminology a recent advance in the methodology of nonlinear models, the “KHB method”. We use data from the NLSY97 to illustrate that existing strategies for estimating the impact of paternal incarceration on son’s delinquency produce biased estimates, and we use high quality registry data from Denmark to show that this is true even across sentence length and across the distribution of sons’ abilities. Our bias-corrected estimates still support the claim that paternal incarceration is damaging to sons’ delinquency, but they also cause concern regarding the widespread use of nonlinear models in criminological research.

Authors

• Lars Højsgaard Andersen, Rockwool Foundation

• Kristian Bernt Karlson, University of Copenhagen