For a given test software project and a random group of developpers,
a single sentence in the specs estimating that it should take 2 or 20 months to complete,
is the main statistical drive explaining why it took, in the end, 5 or 15 months to produce it,
regardless of how much experience the developpers have in the domain,
how much experience they have in software development in general,
what techniques they're using or how familiar they are with the tools.

This is known as the Anchoring effect bias

Source: Greg Wilson - What We Actually Know About Software Development, and Why We Believe It's True @ 15:55