> I am very often able to trade an increase in clock time elapsed (taking more breaks) for a decrease in working time, and vice versa.

But this is the problem with a survey of "how long does the homework take?". The question isn't well defined. The answer doesn't exist.