Monday, September 01, 2008

Correlation is not Causation

There is a fundamental difference between correlation, two or more things appearing together, and causation, item A causing item B to occur. As shown by Levitt and Dubner, there are many examples in public policy where items correlate but there is no causative relationship between the variables despite common acceptance that there must be a relationship, such as homes with books in them and academic achievement (correlation but no causation).

But surely in science, causation can be identified, related mathematically and there can be no confusion between correlation and causation: right? Well, yes -- except where the science does not show a causative relationship, one can not be demonstrated in the language of science (mathematics) and indeed, what we have is not causation, but correlation.

Examples abound in environmentalism, but the most commonly asserted relationship that is not causative is that between temperature increase and anthropogenic greenhouse gasses: the whole theory of AGW rests on the presumption that there is indeed a causative relationship between greenhouse gas emissions and temperature change.

Well, that's easy, everyone knows that there is scientific consensus on this. O.K., if that's the case, then citing a refereed journal publication that shows this causative relationship should be simple. Unless of course, there isn't one. Which is the gist of a series of posts by Jennifer Marohasy who posted a request on a couple of blogs requesting citations of research papers that show a causative basis for AGW.
  • There are of course the voluminous reports from the United Nation's Intergovernmental Panel on Climate Change, with their findings and theories on popular Anthropogenic Global Warming (AGW) theory. The content of these reports, endorsed by governments around the world, have been repeated over and over, for example, in the recent influential report by economist Ross Garnaut to the Australian government.
  • It is apparent, however, that a body of science published in peer-review journals, establishing a causal link between anthropogenic carbon dioxide and warming and quantifying the extent of this warming, is lacking but would be expected to exist to support popular AGW theory.
Asserting that a relationship exists is not the same as substantiating one. The IPCC was not established to see if AGW exists: its mandate is to examine what is to be done about AGW -- a causative relationship is presumptive within the IPCC mandate.

Again, if something is so basic and so clear, providing a concise citation for supportive documentation should not be that taxing.