Correlation is not Causation but it sure is a hint
Correlation is not Causation but it sure is a hint
Edward Tufte (2003)
Pummeling Connecticut with Being-Last-in-the-Latest-Competitiveness-Index is not fun anymore. It’s too easy. Yet it was important to point out why it was so easy. My last post alluded to the nature of index construction. I showed and concluded that the Index construction business was a self-serving exercise (often) and thus arti-factual; that is to say, the rankings suggested by the Indexes were product of the construction method utilized.
This criticism of index number construction is well known: I cited two recent papers (Fisher 2013; Kolko, Neumark and Mejia, 2013).
Fisher and Kolko, et al., successfully impugn and impeach the Indexes they scrutinize and the associated, derived state rankings. In fact, they conclude that the indexes are subjectively constructed, technically suspect, and assembled in a self-serving manner. And I agree.
But they cannot avoid a serious problem.
Fisher & Kolko,et al. not so much err when they conclude that the Indexes are misleading or incorrect. Rather they appear unable to avoiding committing the opposite fallacy – in their subsequent discussion. They fall all too easily into the Rumsfeldian “Absence of Evidence is Not Evidence of Absence” trap. Or its close variant, “Correlation is Not Causation” Trap.
In other words, they dismiss the observed correlation between economics variables and poor economic outcomes as not proving causation and therefore meaning-less.
Similarly, after successfully impugning the index, Fisher & Kolko, et al. conclude that because the flawed Index does not support the Index purveyors point: that Connecticut is up a creek - as proof that therefore Connecticut is NOT up a creek.
We have heard this in other forums: “no sir, the documented correlation between smoking and the incidence cancer is not troubling because everyone knows that correlation is not evidence of causation.”
But it is.
Perhaps what they meant to say – or should have said – is that correlation is not PROOF of causation. But it certainly evidence.
Here is an expert from the New York Times piece by Errol Morris, “The Certainty of Donald Rumsfeld:”
ERROL MORRIS: He used it again and again. He used it in the case of weapons of mass destruction in Iraq. Presumably, the argument is, “Just because we can’t find any evidence that Saddam Hussein has weapons of mass destruction, that doesn’t mean he doesn’t have weapons of mass destruction.”