Meta-analysis of 35 studies examining the effect of indoor temperature on office work performance

Several relationships between air temperature and work performance have been published. We reanalysed the one developed in 2006 by Seppänen et al.; which is probably the best known. We found that even when significant, its prediction accuracy is very low (R² = 0.05, MAE = 1.9%, RMSE = 3.1%). We consequently reviewed the literature and found 35 studies on the effects of temperature on office work performance. We used Seppänen et al.’s approach to normalise the data reported in these studies and explored the feasibility to develop a new relationship using regression models, models based on the Maximal Adaptability framework, and machine learning. We could not find a relationship between temperature and office work performance neither for the range of temperatures measured in most of the office buildings (20 °C–30 °C) or a wider range (18 °C–34 °C). Plausible reasons are discussed including the variety of methods used to assess performance, the multiple uncontrolled confounders, and the fact that temperature alone may not fully describe how the thermal environment affects building occupants. We do not recommend the use in practice of any of the models relating temperature to office work performance examined in the present study. The lack of relationships does not necessarily refute that temperature affects the performance of office work. Coordinated research predicated on a shared protocol enabling integrated analysis in the modelling of the relationships between the indoor thermal environment and office work performance is proposed to be carried out before using them in practice. We made the database open-source and developed an application for data exploring.