Jose Ali Porras-Salazar reanalysed the best-known relationship between air temperature and office work performance, published by Seppänen in 2006. Its predictive accuracy turned out to be very low: an R² of 0.05 and a mean absolute error of 1.9 per cent. The team then reviewed the literature, found 35 studies on temperature and office work, normalised them onto a common footing and tried regression, an adaptability framework and machine learning. None of it produced a relationship, neither across the 20 to 30 °C found in most offices nor across a wider 18 to 34.
Porras-Salazar does not claim temperature has no effect on office work. The claim is narrower: none of the published models should be used in practice, and temperature on its own may not describe what the thermal environment does to people. The laboratory studies on the next sheet find a clear heat decrement, and they run to far higher temperatures than an office reaches. Between 18 and 34 °C, which is the range a workplace actually lives in, the relationship does not hold up.
Joseph Allen changed the air rather than the temperature. Twenty-four people spent six full working days in a controlled office, blinded to the conditions, which simulated a conventional building with high volatile organic compounds, a green building with low ones, a green building with a high outdoor air rate, and artificially raised carbon dioxide. Cognitive scores were 61 per cent higher on the green day and 101 per cent higher on the two green-plus days, at p below 0.0001. Carbon dioxide and volatile compounds each had an independent association.
Rajat Gupta monitored two offices continuously for 19 months, one naturally ventilated and one mechanically. Carbon dioxide was the clearer signal. In the mechanically ventilated office, task scores dropped above 1,000 ppm. In the naturally ventilated one the threshold was 1,400, and median scores below it were, in the paper’s own wording, up to 12 per cent higher. Occupants of the naturally ventilated office tolerated a wider range of conditions. Gupta’s conclusion follows from that: holding an indoor environment inside a narrow band may produce people who cannot cope with anything outside it. The paper reports the two buildings and the 19 months, and not how many occupants sat in them.
Guojian Li put 30 people to work at 25, 27, 29, 31 and 33 °C with everything else held constant. Comfort and performance both fell as it got warmer, sick building symptoms were reported as more intense, and motivation dropped. Best performance came at a thermal sensation vote of −0.13, fractionally on the cool side of neutral. Li’s point is that emotion and motivation carried part of the effect, so the path from the thermostat to the work runs through how someone feels about the room.
Juan Palacios surveyed the same people before and after 70 per cent of a Dutch municipality’s workforce moved into a building designed for health and sustainability. Perceived conditions improved, sick building syndrome symptoms fell, and the mediation analysis traces a path from health and wellbeing to job satisfaction and on to a 2 per cent lower prevalence of sick leave. The effect was still there two years on and was largest among employees over 50. The paper gives the share of the workforce that moved and not the number of people.
Andrea Martina Aegerter counted the working time lost to neck pain among 120 Swiss office workers, across 517 observations. It came to 12 per cent: 1.2 per cent as absence and 10.8 per cent as being at work and impaired. Split by whether someone had neck pain at baseline, it was 14.9 per cent against 0.8. A 12-week programme of exercises, information and ergonomics cut the loss by 2.8 percentage points, at p = 0.049 and a confidence interval that touches zero.
Porras-Salazar pools other people’s data and is also the only study here that reports finding nothing. Allen worked with 24 people over six days in a chamber and Li with 30 in a field office. Gupta measured two buildings for 19 months. Palacios and Aegerter worked with one municipality and two organisations.