Kawakubo measured productivity in a real office with the air conditioning set first to a predicted mean vote of 0.0 and then to −0.5, and analysed the responses with the personal attributes of the workers in the model, which earlier studies had left out. Men were most productive at a thermal sensation of −0.61, between slightly cool and neutral. Women were most productive at +0.23, between neutral and slightly warm. The two optima are 0.84 scale points apart, and the settings the office actually ran at, 0.0 and −0.5, are neither of them.
Thomas Parkinson counted complaints instead, in two independent US datasets: 38,851 responses to the Center for the Built Environment survey from 435 office buildings across 168 cities, and a set of posts crowdsourced from a social platform whose size the paper does not give. Sixty-six per cent of the posts about cold offices came from women and 34 per cent from men, against an estimated 55 per cent of US users of that platform being women. Uncomfortable office temperatures were more likely to be too cold than too hot regardless of season, and the seasonal pattern differed between men and women: the women’s complaints rose in summer, out of phase with the men’s. The vocabulary differed too. Women wrote about sweaters, jackets and blankets; men did not.
Sudimac and Kühn took 63 participants for a one-hour walk, half in a natural environment and half in an urban one, with an fMRI paradigm inducing social stress before and after. Amygdala activity decreased after the walk in nature and it did so only in the women. Performance on the arithmetic tasks improved in the women too. After the urban walk, amygdala activity did not fall in either group.
Bin Jiang put 160 people through a laboratory stress test and then showed each of them one of ten six-minute videos of a neighbourhood street, with tree canopy ranging from 1.7 to 62 per cent, measuring salivary cortisol and skin conductance. For men the curve was an inverted U: recovery improved as canopy rose to 24 per cent, was flat from 24 to 34, and got slower above 34. For women there was no relationship at any density at all.
Wels pooled seven UK longitudinal studies, 34,131 observations across three stages of the pandemic, and related home working to psychological distress, low life satisfaction, poor self-rated health, low social contact and loneliness. The associations differed by sex. This paper is a preprint on medRxiv and has not been through peer review.
Wels’s 34,131 observations and Parkinson’s 38,851 survey responses are the large samples, and both come from data collected for other purposes. Jiang worked with 160 people in a laboratory and Sudimac with 63. Kawakubo surveyed one office and reports no participant count in the abstract.