Treatment effects under endogeneity and non-random selection: Estimating the impact of stress on addictive substance use
Anastasia Gergenreter has published an article «Estimation of treatment effects on ordinal variables in multivariate ordered choice models» in Applied Econometrics . The study offers a fresh perspective on estimating treatment effects for ordinal outcomes in settings where non-random selection is present and the conditional independence assumption no longer holds.
Anastasia Gergenreter has published the article «Estimation of treatment effects on ordinal variables in multivariate ordered choice models» in Applied Econometrics .
The paper addresses the problem of estimating treatment effects when the key assumption of conditional independence between the treatment variable and the outcome is violated. The author proposes an extension of existing econometric approaches through a parametric method based on a hierarchical system of ordered choice equations. The study provides closed-form analytical expressions for the average treatment effects and average marginal effects in the context of multivariate ordered probit models, thereby enabling consistent estimation under conditions of endogenous treatment assignment and non-random sample selection.
An empirical application of the proposed framework is presented in the context of health economics, addressing the effect of stress on the consumption of addictive substances. The analysis draws on NSDUH survey data from the United States covering the period 2020–2023. Using a hierarchical system of equations helps correct the bias that arises from the fact that stress levels are not exogenously determined.
The results indicate a consistent positive association between stress levels and the probability of substance use. The estimated effects further suggest that women exhibit greater susceptibility to stress-related addictive behavior relative to men. A comparison with estimates obtained from models that do not account for endogeneity reveals notable discrepancies, thereby confirming the empirical relevance of the proposed methodological framework.
These findings have important implications for social policy and public health. The observed variation in stress effects across socioeconomic and demographic groups suggests that uniform prevention strategies may be insufficient. The methodological framework proposed in this study allows for more accurate identification of at-risk populations and can inform the design of targeted interventions aimed at reducing addictive behaviors.
