New Framework Combines Statistics and Social Theory to Explain Hidden Causes
Researchers have developed a method that merges two dominant analytical approaches to better understand cause-and-effect relationships in social phenomena—particularly around poverty and neighborhood decline. The work addresses a persistent problem: how to prove causal claims when key factors can't be directly measured, a challenge affecting policy design and program evaluation.
Originaltitel: Navigating Stronger Evidence: Unobservable Processes and Causal Mechanisms in Social Trajectories
<p>This dissertation explores the interplay between statistical causal inference and mechanism-based explanations in analytical sociology, focusing on the challenge of unobservable processes and their implications for robust causal explanations. Analytical sociology emphasizes the importance of mechanisms in understanding social phenomena but often grapples with the integration of statistical causal inference methods. By combining two prominent frameworks—the structural causal model (SCM) and the potential outcomes approach (PO)—this work bridges gaps in addressing unmeasured confounders and alternative mechanisms. It contributes to the advancement of theory-driven causal inference by emphasizing evidential pluralism and triangulation, integrating diverse forms of evidence to enhance the validity of causal claims.</p><p>The dissertation comprises three empirical papers, each addressing a critical research question. Paper 1 investigates the mediating role of disadvantaged neighborhoods in social assistance dependency in Sweden. Using Swedish register data, it identifies pathways linking early social assistance usage, neighborhood disadvantage, and long-term dependency. Results indicate that disadvantaged neighborhoods play a central mediating role, especially for individuals at high risk of early social assistance, highlighting the interplay of contextual and individual-level mechanisms. Paper 2 examines the relationship between homeownership and unemployment, exploring temporal dynamics and effect heterogeneity across migration backgrounds, education levels, and neighborhood characteristics. Employing advanced causal inference methods, the study finds that homeownership provides protective effects against unemployment, with stronger effects observed for immigrant homeowners. These findings underscore the nuanced, context-specific mechanisms underlying the homeownership-unemployment nexus. Paper 3 addresses the role of unmeasured confounders in neighborhood selection and broader sociological research. Through a scoping review and an expanded sensitivity analysis framework, this paper demonstrates the importance of assessing potential biases and alternative explanations in causal studies. It provides methodological guidance for evaluating the impact of unobservable processes across different empirical settings.</p><p>Together, these papers as main cases advance analytical sociology by integrating statistical causal inference and mechanistic reasoning to provide robust, theory-driven explanations of complex social phenomena. By addressing unobservable processes and emphasizing evidential pluralism, this dissertation contributes to the development of more comprehensive and context-sensitive sociological insights.</p>