Within a freight transport context, the origin-destination distance and the weight of the shipment play an important role in the decision of the most preferred transport service and in the way logistics managers evaluate the transport service's attributes. In particular, the attributes commonly used in order to describe a freight transport service in a stated choice framework are cost, time, punctuality and risk of damages, respectively. This paper investigates the role of origin-destination distance and weight of freight transport services introducing a conditioning effect, where the standard utility function is conditioned on the freight transport distance. We refer to this model form as a heteroskedastic panel multinomial logit (panel HMNL) model. This model form outperforms the underlying unconditioned model and suggests that an appropriate conditioning effect leads to an improved understanding of the derived measures, such as measures for marginal rates of substitution.