Lane changing behaviour occurs frequently on the highways. However, it also poses a major impact on traffic operation and safety since complex interactions between two or more vehicles on different traffic lanes are involved. In the lane-changing related crashes, correlation in damage level among the vehicles involved is prevalent. To this end, a copula approach is proposed to model the vehicle damage level of lane-changing related crash, with which the dependency between lane-changing and lane-keeping vehicles is accounted for. Additionally, a semi-parameter estimation approach is adopted to address the problem of heterogeneous data structure. In this study, crash data from Orlando City of Florida during the period between 2016 and 2019 are used. Then, the semi-parameter copula-based ordered logit models are estimated to measure the association between road environment, vehicle attributes, driver characteristics, crash circumstances, and vehicle damage level of two-vehicle lane-changing related crashes. Results indicate that there are major discrepancies in the influences of possible factors on vehicle damage level between lane-changing and lane-keeping vehicles. Furthermore, non-linear relationships between vehicle damage level, driver age, and time of crash are also revealed.