In congestion management, total traffic delay is typically derived based on look-up tables for incident duration, capacity reduction, and frequency that are determinants of total incident delay. The studies that adopted this approach indicate that some factors, such as weather, may not be included in the look-up tables because of their limited size and the lack of rigorous identification of influencing factors to be included in the tables. In addition, the marginal effect of the influencing factors on total traffic delay cannot be analyzed. In this study, statistical models with identified influencing factors for incident duration, lane blockage, and frequency were developed and then integrated into the queuing-based delay equation. Through the development of advanced statistical models, the variables or factors that significantly influence incident duration, lane blockage, and frequency can be identified. When these statistical models are plugged into the queuing-based delay equation, the marginal effect of the identified factors on total delay can then be derived analytically. On the basis of incident data collected in New York City, the proposed approach was demonstrated to be successful.