The debate over the problems in American education and some of the proposed solutions, such as merit pay and easy teacher termination, reminds me of one of the cardinal sins of educational statistics: correlation = causation.
Just because two things seem to be related, doesn't mean that one caused the other. You know: the rooster crows and the sun rises; therefore the rooster causes the sun to rise. Four out of five gang members chew gum; therefore a ban on chewing gum will eliminate gang violence. School A does X and is judged successful; School B does Y and is judged unsuccessful. So, forcing School B to do X will make school B successful.
Except, of course, School B often gets worse.