Title: Behavioral Decision-Making and Asymmetric Herding in Organizational Emergency Evacuations: A Drills-Based Empirical Study Abstract: This paperr examines individual and social decision-making processes during unannounced fire evacuation drills within an organizational setting. Through a controlled field experiment involving 113 graduate students in an office building, post-drill survey data were analyzed to model evacuation intentions and herding behavior using modified expected utility and nonlinear conformity frameworks. Results indicate that traditional expected utility theory inadequately explains evacuation intention, which instead follows a threshold-based, risk-averse pattern. Furthermore, herding effects were found to be asymmetric: individuals initially inclined not to evacuate were significantly more likely to conform to evacuating peers than vice versa. These behavioral insights were formalized into piecewise utility and quadratic herding models, validated via agent-based simulation in NetLogo. The findings contribute to emergency management and organizational behavior literature by quantifying decision dynamics under threat, supporting the design of more effective evacuation protocols, training programs, and risk communication strategies tailored to group behavior in workplace emergencies.