Metodologia de análise dos determinantes da demanda por transportes no paradigma da inferência causal

Abstract

Despite the emergence of behavioral modeling in research efforts to explain travel demand, questions still arise as to the causal interpretation of the estimated coefficients, with endogeneity being a source of bias that leads to the distinction between causality and correlation. In this master's research, we propose a method that seeks to contribute to the investigation of causal relationships in travel behavior, incorporating the theoretical framework of Causal Inference as a tool that formalizes criteria for the identification of causal effects and allows structuring statistical tests on the causality hypotheses. Through a systematic review of the literature, a conceptual representation of travel behavior was consolidated, which highlights the central role of accessibility in explaining mobility patterns. Then, aggregating the theoretical basis of causal inference, we sought to identify gaps in current transport demand modeling efforts, showing that limitations in the incorporation of control variables can explain, in part, the different conclusions obtained in empirical studies on the determinants of travel decisions. Finally, the previous conceptual and methodological bases have subsidized the development of a method that allows for the formulation and formal testing of causality hypotheses through Structural Equation Models. The obtained results highlighted the applicability of the proposed method, which brought methodological and phenomenological contributions with important implications in behavioral modeling and in the transport planning process.