Wednesday, September 28, 2005

Blaming The Tiny Dwarf

There is a classic Saturday Night Live sketch wherein Steve Martin plays the Medieval Barber, who is doing his best "barbering" for a variety of peasant ailments. When someone questions the wisdom of one of his "cures," Martin looks at them shakes his head and says "Who's the barber here?" At another point Martin's character pats himself on the back for the vast strides his profession had taken from years past. To a woman concerned about her daughter's illness he states, "In the past we would have thought that your daughters condition meant she was possessed by demons. [laughs] But, now we know her condition is caused by a tiny dwarf living in her stomach."

This sketch was brought to mind after I read this scandalously misleading article in Time about hurricans and global warming. There is so much wrong with the article that a full scale fisking is beyond me. (I simply do not have THAT much time.) But read the thing yourself with a few of these points in mind.

A) How well does the theory presented stand-up to the generally accepted methods of scientific discovery? One example, do they have good sample sizes for the data they are looking at? Is basing your claims on 10 years of hurricane data acceptable? How about 20? How about 50?

Another example: Are they making claims that are either I) testable, or II) falsifiable? Making claims that Katrina and Rita prove storms are "more intense" than storms from previous decades because they were at sea large Category 5 storms, although they struck land at Category 3 intensity, is a spurious claim. We have no idea what the intensity levels were at sea for the thousands of Atlantic hurricanes from 1850-1960. Therefore, there is nothing to compare these storms to except the last 20-30 years worth of data. (And even in the Time article a meterologist insists that satellite data until 1989(!) is suspect. I'm not sure I buy that, but if even this is an open question, how much "confidence" should one place in these definitive statements?) No one can either prove or disprove the intensity levels ascribed to pre-1960 hurricanes while they were still at sea. Therefore a claim that storms are more intense now is neither testable nor falsifiable, barring access to a time machine of course.

B) Are they cherrypicking data? You have to ask yourself this especially if you are not up on all of the data involved. For example, from the Time article:

On the whole, they found, the number of Category 1, 2 and 3 storms has fallen slightly, while the number of Categories 4 and 5 storms--the most powerful ones--has climbed dramatically. In the 1970s, there were an average of 10 Category 4 and 5 hurricanes a year worldwide. Since 1990, the annual number has nearly doubled, to 18.

This would all be fine and dandy if the 1970's were a representative decade. It very clearly isn't. Since 1870 (!) no decade, except the 1980's, had fewer hurricanes in general or fewer large hurricanes in particular than the 1970's. (See this NOAA data, for example.) But even aside for that picking one decade (say the 1970's) and saying "this is normal, any deviation from this is abnormal" is ludicrous. Look at ALL the data and you won't wind up making false comparisons. However, they don't want you to look at all of the data because it doesn't support their findings. For example this is the data for the entire Atlantic Basin from 1950 to 2000:

Decade-Tropical Storms-Hurricanes-Hurricanes As % of Storms-# Large Storms (Cat 3+)

1950's - 104 - 60 - 57.7% - 13
1960's - 95 - 50 - 52.6% - 14
1970's - 95 - 38 - 40.0% - 5
1980's - 93 - 37 - 39.8% - 7
1990's - 111 - 50 - 45.0% - 13

The choice of the 1970's, in this case, seems simply dishonest, doesn't it?

C) Are the claims internally consistent? The claim is but forward that we are witnessing sudden dramatic change of an unprecedented nature. However, at the same time any conflicting data is explained away as being within the parameters of "normal" climitalogical conditions. Well, which one is it? How can you have "huge unprecedented changes" that result in "normal climitalogical conditions"? I'm not saying that it cannot work that way, I am saying there is no reason to assume it works that way. Show me WHY warmer water will not help tropical depressions and lows become category 1 or 2 hurricanes but will help category 2 or 3 hurricanes become 4 or 5's. Don't just say, "Oh we have fewer hurricanes now because that is natural."

D) This brings me to the point of postulating unknown forces to back up your theory. A good rule of thumb is never subscribe to any theory that relies on such a rhetorical device. From the Time article:

So that ought to mean a lot more hurricanes, right? Actually, no--which is one of the reasons it's so hard to pin these trends down. The past 10 stormy years in the North Atlantic were preceded by many very quiet ones--all occurring at the same time that global temperatures were marching upward. Worldwide, there's a sort of equilibrium. When the number of storms in the North Atlantic increases, there is usually a corresponding fall in the number of storms in, say, the North Pacific. Over the course of a year, the variations tend to cancel one another out.

Notice how all of this is contigent upon the existence of something we might call the "Hurricane Equilibrium." The trouble is we don't know if it exists. (Not to mention the sticky problem of correlation v. causation implied in the scenario.) So, in order not to have to deal with conflicting information (i.e. the relative paucity of actual hurricanes while conditions are ripe for starting them) we are supposed to believe that this unknowable "Equilibrium" exists AND that instead of the "Equilibrium" being affected by by the "sudden unprecedented change" of global warming it is in fact functioning normally and regulating the number of hurricanes. I hear that and I can only think they are blaming the dwarf instead of the demon.


dadvocate said...

Good analysis. Sample size is one of the critical factors here and it is necessarily small because of the lack of data.

The earth has been warming up for thousands of years. Sheets of ice used to reach what is now the Ohio River. Something made the climate begin to get warmer. I doubt it was anything human related.

There are also a large number of complexly interrelated variables. It seems premature to be pinning the blame on human impact as strongly as many do.

I made a similar analysis of the hurricane data here a couple of weeks ago.

The Iconic Midwesterner said...

I like your analysis as well. It certainly gets to the heart of the "intensity" question. I'll point folks here to it from a post later tonight.

Pursuit said...

Well done. There is an article in the Times today talking about the Arctic sheet ice. In the same paragraph they at first say that the ice has not receded this much in one hundred years. Then they say it is hard to not see the cause as human caused global warming. The necessary implication to a rational person would then be that the recession of the ice a century ago was also human caused global warming, which is of course ridiculous. In there zeal to sell us on their hysterical position, they completely miss this point.