Thursday, February 4, 2016

Confirmation

I have long argued that many of our supposed crises were nothing more than statistical artifacts. For example, though childhood obesity has increased, the problems with obesity in general, and childhood obesity, have been made to appear much more sudden and severe by changes in definitions of what constitutes obesity. (See "Twice in a Row".) The same is true for the sudden increase in autism. As I argued in "Statistical Artifacts", while many seek a cause in various places, in truth, much of the crisis is simply the result of changing definitions and greater efforts at diagnosis. The same is true for several other problems, such as depression and childhood learning disorders, where media attention, shifts in diagnostic criteria, and efforts to improve diagnosis have given the appearance of increasing numbers where, in truth, we simply are seeing the results of changes in definition and efforts at diagnosis.

Nor is this problem limited to such simple outcomes. For example, I have pointed out (see "Too Much Knowledge", "Funny Numbers" and "Shocking Numbers") that our improving ability to find chemicals, moving from parts per million, to parts per billion or even trillion, has allowed us to find traces of chemicals where before we found none. But, rather than recognizing this is simply a change in technology, we are told by the media that this is evidence of increasing chemical contamination, an utterly absurd allegation.

Similarly, thanks to the media covering child abductions due to the need to fill the 24 hour media cycle, we now hear of abductions in far away states where previously we heard only of those few are incidents of local abductions. As a result of increasing reports, and reports coming from father away, we have come to imagine stranger abductions are a crisis, where in truth they remain very rare, and the incidence is not increasing.

Even environmental issues have suffered from this conflation of reporting with reality. As I argued in "The World's Most Stupid Bureaucrat" and "Confirmation, Again", we often confuse monetary values with absolute magnitude. For example, where, in the past, a hurricane would have ravaged an empty coastline, new development now means thousands die and millions of dollars of damage are done. When we hear of this, rather than recognizing that changing patterns of development have made storms more damaging, we often hear about the increase of storm magnitude and blame it on climate change. But, in truth, the storm is no worse, there are simply now people there to be harmed -- and to report on the storm -- where previously it would have passed unrecognized. Likewise, where past storms would have felled trees and shifted rocks, they now tear down buildings, and the "damages" are reported as rising, while in truth it is not the storm that increased in magnitude, but rather the value of the potential targets. (See also "The Rubber Yardstick".)

And then there are even less obvious faux crises. For example, the increase in deaths due to cancer and heart disease. While, in one sense, these may be worrisome, especially if coming at younger ages, in another, they should be reassuring. Everyone will die of something. Cancer and heart disease are largely deaths of comfortable old age. When we were dying of disease, animal attacks and starvation, we did not live longer enough to develop cancer or heart disease. The fact that we now live long enough for our bodies to begin to fail should be a cause for celebration, not worry over cancer and heart disease. (Cf "Dubious Claims". See also "Lifespan" and "Poverty and Lifespan", for a discussion of another abuse of numbers.)

As should be obvious, I have been making this case for a long time, in many contexts, so it is nice to finally hear some confirmation. And confirmation I did receive. Researchers in Denmark have discovered that, by and large, the supposed crisis in autism was largely the outcome of changing diagnostic criteria and inclusion of outpatient as well as inpatient diagnoses, at least in the countries they examined.

Unfortunately, I doubt many will believe this evidence. Far too many have an emotional stake in these crises, and other have a financial interest. And thus, though the evidence is clear for those who care to see it, I doubt it will make much difference.

POSTSCRIPT

One other nice bit of support is the argument in the essay that causation and correlation are not the same, using the example of charts showing autism growth and growth in organic food consumption are almost perfectly correlated. I have made this argument in "Correlation vs. Causation" and "Correlation and Causation Revisted", though using different examples. Still, nice to hear others pointing out this simple, but often overlooked, truth.

No comments:

Post a Comment