It seems obvious that the solutions we apply to populations are different than those we apply to individuals. There are plenty of individuals who can safely travel at greater than the posted speed limit but the population, generally, would be less safe if everyone could. Everyone believes themselves to be the individual who can safely be the exception, thereby making it impossible to have exceptions. Additionally, the interactions of different individuals can make a situation that is safe for either independently, unsafe for them together.
There is an article making the rounds (seemingly everywhere) about how a study has found that there is no safe amount of alcohol. Many of the articles I have read about this topic struggle to rationalize this outcome against the many, many studies that have found moderate amounts of alcohol can improve the health of individuals. Studies of populations are very different from studies of individuals to the point you cannot compare the results.
The same situation exists in workplace design. An office designed for how a particular person works best will likely not work well for the typical employee. The best designs seek to allow the most people possible to reach peak productivity. You can never make a space perfect for everyone.
You can draw high-level understandings of populations by measuring the behavior of individuals. But much like the behavior of a single driver is dramatically different than the behavior of a pack of cars on the interstate, individuals describing their workday are dramatically different than how groups are observed to work together in offices.