Compared to average outdoor levels.
"Significant" is subjective here, but it needs to be a large enough effect that people would actually care to fix their air quality. (For example, asbestos has a "significant" health risk.)
Resolves once there's a well-established scientific consensus.
Assuming the research was done well, then it's a question of is the effect large enough to care. Does a 1% increase in test response time and 4% increase in test throughput translate into, say, a 1% increase in productivity for well-compensated knowledge workers?
@IsaacKing Responses per minute on the Stroop test, the test where a color word is printed in some color that does not match the word and you need to state the color the word is printed in.
@WesleyJB Does accuracy stay the same? If so, if that replicates I think that's significant. Companies would definitely care in a 4% increase in worker speed.