Displaced populations are housed in various constructions, including lightweight predesigned structures. Theoretically, self-built heavyweight structures should ensure better temperatures and be closer to cultural norms. To examine this directly for the first time, lightweight pre-designed solutions are compared with high-mass self-built alternatives in Afghanistan and Peru, via monitoring, dynamic simulation, occupant surveys, the Shelter Assessment Matrix (SAM) and ShelTherm. Lightweight solutions increase peak summer temperatures, but only by 2°C, but reduce minimum temperatures by up to 5°C. Simulations only provided a qualitatively similar time series to the monitoring, because identical homes showed a large variance in temperatures. This questions the benefit of simulation compared to approaches which concentrate on whether shelters exacerbate or ameliorate external temperatures. In addition, a dwelling provides more than comfort, it supports family life, which is best addressed by tools like SAM, not thermal simulation. Hence it might be ideal to recommend high-mass self-build if possible, and to focus modelling efforts on qualitative aspects of simulation time-series, such as whether the building suppresses or exacerbates external conditions, and equally on psycho-cultural aspects. The term narrative modelling is introduced to describe this new approach which will be of direct benefit to the humanitarian community.