Abstract for: Estimating Information Delays in Risk Perception: A Hybrid ARDL-Erlang Approach to Modeling Mobility Change During a Pandemic

Accounting for the time needed to acquire and process information is crucial when modeling risk perception. Estimating the average delay period and order (distribution shape) is necessary but challenging without direct measurements. Econometric methods, such as Autoregressive Distributed Lag (ARDL) models, provide empirical rigor but lack theoretical clarity for simulations. In contrast, system dynamics models use structured delay distributions like Erlang lags but struggle with empirical estimation. Our method combines econometric estimation with system dynamics modeling. We extract delayed behavioral responses from empirical data using ARDL models and translate these estimates into theoretically transparent Erlang delay structures. This transformation captures three characteristics: total impact, average delay length, and delay order (shape). We apply this method to mobility responses during the COVID-19 pandemic, estimating how delays in processing information about deaths influenced county-level mobility across the United States. Our analysis demonstrates the applicability of the proposed method and key findings on pandemic-related mobility shifts. Nationally, information delay follows a 1st-order Erlang distribution with an average delay of 3.2 weeks. However, delays vary across states, with estimated delay structures ranging from 1st to 3rd order and average delays spanning 2 to 18 weeks. These findings highlight variability in how populations process risk-related information and modify their behavior. By bridging econometric estimation with system dynamics delay structures, this study provides a novel approach for quantifying and interpreting delayed human responses. Our method enhances theoretical transparency in behavioral modeling and facilitates the integration of empirical estimates into simulation models for public health and policy analysis. More broadly, this framework offers a replicable approach for identifying and modeling information delays in other human-environment systems.