Modelling lane changing behaviour in approaches to roadworks: Contrasting and combining driving simulator data with stated choice data

Drivers approaching lane closures due to roadworks tend to choose a target lane (plan) and seek suitable gaps to execute the plan (action). The plan is however latent or unobserved as the driver may or may not be able to move to the target lane due to the constraints imposed by the surrounding traffic. Hence, only the actions of the driver (as manifested by their final lane occupancies) are observed in the trajectory data. This paper analyses such mandatory lane changing behaviour in a roadworks environment in detail with data from a controlled driving simulator experiment and a simple stated preference survey with the same group of participants. While in the former drivers face similar constraints in implementing the plans as in the real world, in the simple stated choice survey the same drivers elicit their preferred target lanes without a need to put the plan into action. We contrast the findings from the two sources and also show correlations between the latent plan and stated target components in a latent class model. The results provide new insights into lane changing behaviour that may be useful for example for traffic management purposes. Furthermore, using stated choice data potentially reduces the cost of data collection for model development.