As a general rule, people who believe in a particular philosophy search for data that supports that philosophy and discard data that does not. Someone who wants to convince others of a position obviously uses supportive, not unsupportive, data. When you read, "Use method X. You too can achieve up to 93 percent increases in productivity (or quality)," the method may really have achieved such results. But it was probably the exceptional case. In all likelihood, most projects experience far less dramatic results. And some projects may even experience decreased productivity using method X.
Reference:
Fenton, N., "How Effective Are Software Engineering Methods?" Journal of Systems and Software, August 1993.
