[criticaltinker]

> How can it not be causality when it's a double-blind intervention?

The design of the study looks solid and results are certainly statistically significant. With N=24 and the limitations quoted below, I think it's fair to take these results with a grain of skepticism.

But on balance, I found the criticism in GP's comment to be a little over the top.

- "By testing on subsequent days, it is possible that effects from one condition were reflected in the scores obtained on the next day."

- "The environmental factors that were not experimentally modified exhibited some variability owing to changes in outdoor conditions and participant behavior."

- "This study used a controlled environment to individually control certain contaminants. Assessments performed in actual office environments are important to confirm the findings in a noncontrolled setting."

[fighterpilot]

It's not just N=24 because those same participants took 9 independent tests which generates more data. Also, the p-value should be considered along with the N, since the sampling variance of the test statistic will go up with a small N and this is accounted for in the test thresholds, which somewhat offsets that limitation.

But, a larger N and replication studies are needed.

[criticaltinker]

Good points, one question - did you mean to say "6 independent tests", or am I missing something?

[fighterpilot]

I thought it was 9, but I might have misread the study.