Doors to Productivity [...]

Unusually, there is actually experimental research on this issue. (Mostly, assertions of best practices can never be backed up with good research, because it would be too expensive.) Back in the 1970s and 1980s Barry Boehm had enough programmers working for him, and enough money, to do meaningful experiments. He wrote up the results in one of those classic bricks of computing books, Software Engineering Economics 1981, and still in print. He found, among other things, private offices (with doors that shut) increased productivity about 25% over similar programmers doing the same work in the normal cruddy environment. This was the best way to increase programmer productivity by some distance. (Source)