Planning agencies need to assess the change in travel times and diversion rates attributed to work zones to reduce the adverse effect on mobility of the roadways. To assess such changes, simulation-based models can be calibrated with empirical observations to depict traveler behavior in selecting alternate routes. In this paper, Bluetooth detectors were used to capture and model driver route choices in the presence of a work zone. A method was developed to estimate change in travel time and total volume at different routes due to work zones. The proposed method can help identify outliers and erroneous MAC IDs. Moreover, the method is robust in handling low sample sizes and bimodal distributions of travel time. The developed method was applied to a large-scale work zone stretching 8.5 miles of interstate I-40 in Raleigh, NC. Nine Bluetooth detectors were deployed in two rounds — one without work zones and one with work zone conditions. The locations of the detectors were selected a way that captured both work zone routes and alternate routes for each given origin-destination pair. In each round, data was collected continuously for 14 days for different routes, with each route contained three detectors. For three-point routes, change in travel time and traffic volume due to work zone was estimated corresponding two-point routes data. The results showed that travel time increased in work zone routes during PM peak and overnight hours. Consistency in detector placement and alignment was found as an important factor in accurately determining diversion sensitivity.