Modeling anger and aggressive driving behavior in a dynamic choice-latent variable model

Area of Science:

• Psychology
• Transportation Engineering
• Data Science

Background:

• Aggressive driving is a significant road safety concern.
• Existing models often fail to capture the dynamic nature of driving behavior and anger.
• The state-trait anger theory provides a framework for understanding anger in driving contexts.

Purpose of the Study:

• To develop and validate a hybrid choice-latent variable and Hidden Markov model for analyzing aggressive driving.
• To model trait driving anger as a latent variable and state anger as a dynamic latent variable.
• To forecast aggressive driving manifestations and test interventions for road safety improvement.

Main Methods:

• A hybrid choice-latent variable model integrated with a Hidden Markov model was developed.
• Trait driving anger was modeled as a latent variable influenced by individual characteristics.
• State anger was modeled as a dynamic latent variable affected by trait anger, frustrating events, and contextual variables.

Main Results:

• The proposed model demonstrated a superior goodness of fit compared to a simpler joint model.
• State anger was significantly influenced by frustrating events, trait anger, and prior state anger.
• The model successfully captured the dynamic evolution of anger and its impact on driving behavior.

Conclusions:

• The developed hybrid model offers a robust framework for analyzing and predicting aggressive driving.
• This approach can inform the development of effective road safety interventions.
• The model's incorporation into micro-simulation packages can enhance traffic safety research.