In this paper we structure biological, psychological and background/experience drivers of thermal comfort variance and their relationships to develop a conceptual interaction model. The aim is to create a theoretical model containing a broad range of influencing factors that can be used for hypothesis generation. Furthermore, the paper provides a framework for assessing how much of the observed diversity in comfort votes may arise from imprecise instruments through the assessment of various forms of validity. Current comfort models, both predictive and adaptive, focus on the prediction of conditions comfortable for an average person in order to derive comfort bands suitable for the majority of building occupants. Such models do not explain why we observe such a diversity of comfort votes from occupants of the same building. We argue that understanding diversity is important, both practically and scientifically, and that to do so we need to address the physiological, psychological, social, cultural and built-environmental conditions that give rise to observed diversity in comfort. It is expected that in doing so, the research community will both improve its scientific understanding of comfort, but also develop new ways of providing comfort that can create acceptable environments for more people using less energy.