In conventional transmission electron microscopy, thermal scattering significantly affects the image contrast. It has been suggested that not accounting for this correctly is the main cause of the Stobbs factor, the ubiquitous, large contrast mismatch found between theory and experiment. In the case where a hard aperture is applied, we show that previous conclusions drawn from work using bright field scanning transmission electron microscopy and invoking the principle of reciprocity are reliable in the presence of thermal scattering. In the aperture-free case it has been suggested that even the most sophisticated mathematical models for thermal diffuse scattering lack in their numerical implementation, specifically that there may be issues in sampling, including that of the contrast transfer function of the objective lens. We show that these concerns can be satisfactorily overcome with modest computing resources; thermal scattering can be modelled accurately enough for the purpose of making quantitative comparison between simulation and experiment. Spatial incoherence of the source is also investigated. Neglect or inadequate handling of thermal scattering in simulation can have an appreciable effect on the predicted contrast and can be a significant contribution to the Stobbs factor problem.