When Statistics Are MAGIC

Whether judging the worth of someone else’s statistical work or your own, it’s always helpful to keep in mind the “MAGIC” criteria spelled out by Robert Abelson in Statistics As Principled Argument.  Here are the five:

"1.  Magnitude:  How big is the effect?"  Statistically significant or not, is the size of the effect too trivial to matter?
"2.  Articulation:  How precisely stated is it?"  Is the finding too vague or muddled to be useful?
"3.  Generality:  How widely does it apply?"  Does it only matter for one city, one college major, one health condition?
"4.  Interest:  How interesting is it?"  Will it get anyone’s attention?
"5.  Credibility:  How believable is it?"  Not that counter-intuitive findings should be ignored.  But they should be especially questioned:  extraordinary claims require extraordinary evidence.

These five criteria are worth keeping close at hand to help you decide when a statistical finding is really actionable.
​
Contact:  [email protected]
​