I had the knife at my throat after reading a paper by Preti, Lentini, and Maugeri in the Journal of Affective Disorders (2007 (102), pp 19-25; thanks to Marc Morano for the link to World Climate Report where this work was originally reported). The study had me so depressed that I seriously thought of ending it all.
Before I tell you what the title of their paper is, take a look at these two pictures:
The first is the yearly mean temperature from 1974 to 2003 in Italy: perhaps a slight decrease to 1980-ish, increasing after that. The second pictures are the suicide rates for men (top) and women (bottom) over the same time period. Ignore the solid line on the suicide plots for a moment and answer this question: what do these two sets of numbers, temperature and suicide, have to do with one another?
If you answered “nothing,” then you are not qualified to be a peer-reviewed researcher in the all-important field of global warming risk research. By failing to see any correlation, you have proven yourself unimaginative and politically naive.
Crack researchers Preti and his pals, on the other hand, were able to look at this same data and proclaim nothing less than “Global warming possibly liked to an enhanced risk of suicide.” (Thanks to BufordP at FreeRepublic for the link to the on-line version of the paper.)
How did they do it, you ask? How, when the data look absolutely unrelated, were they able to show a concatenation? Simple: by cheating. I’m going to tell you how they did it later, but how—and why—they got away with it is another matter. It is the fact that they didn’t get caught which fills me with despair and gives rise to my suicidal thoughts.
Why were they allowed to publish? People—and journal editors are in that class—are evidently so hungry for a fright, so eager to learn that their worst fears of global warming are being realized, that they will accept nearly any evidence which corroborates this desire, even if this evidence is transparently ridiculous, as it is here. Every generation has its fads and fallacies, and the evil supposed to be caused by global warming is our fixation.
Below, is how they cheated. The subject is somewhat technical, so don’t bother unless you want particulars. I will go into some detail because it is important to understand just how bad something can be but still pass for “peer-reviewed scientific research.” Let me say first that if one of my students tried handing in a paper like Preti et alia’s, I’d gently ask, “Weren’t you listening to anything I said the entire semester!”
Any method that attempts to correlate the two raw data series will end in grief, for the obvious reason that the pictures look nothing like one another. Temperature more or less increases and the suicide rates peak—at different points for males and females—then trail off. The authors never bother to hazard a guess why this is so. But I can ask, is it a coincidence that the highest male suicide rate coincided with Italy’s loss to Brazil in 1994?
Anyway, because the raw series are obviously unrelated, and the authors need to have a paper showing a global warming connection, they have to manipulate the data in some way. The method our authors applied was this: They first fit a trend model to the temperature series (a line), and a quadratic polynomial to the suicide series. Why quadratic? Just looking at the curved, solid line on the plots shows that the quadratic fits terribly, especially for women. But, like the authors, let’s ignore that.
The fact is that no model is needed for either series because these data are the actual data. There is nothing hidden here, nothing that remains uncertain that we need to model with a probability distribution. It’s true that we can model the relationship between the series, but that’s not what they did. They modeled each series by its lonesome, and they modeled it badly.
This is an important point, but a minor one, because their goal was not to actually model the series after all, but to produce the “residuals” from such models. Residuals are the numerical difference between the actual series and its model. The authors then tried to model the residuals from each series to see if these were related. No dice. So they had to manipulate the data further to produce the result they wanted.
What trick did they next try? A “Gaussian low-pass filtering procedure” mainly used in eliminating noise from images (they never show nor explain their algorithm nor attempt to justify its use). You might think of this as a way to smooth the residuals. I have elsewhere on this blog showed how smoothing (creating running means is a smoothing method) can induce correlation between two smoothed series even when those series are absolutely unrelated. This means our authors are more likely to see an effect that is not there. Did they?
No, blast it! Even after all this, they still could not find a publishable p-value connection between the two twice-modified series. But they still had to—had to—prove their theory correct. So what did they do? Anybody?
Yes, they manipulated the series a third time! The trick they used was the oldest in the book: they threw out the data that did not fit their preconceptions! Think I’m joking? Here it is in their own words (recall that all the other manipulations have already taken place):
[W]e were able to identify deviant points (or statistical outliers) in our series; by excluding deviant points (no more than 1-2 per series), a sensitive increment in the association between our measure of temperature and suicide (Table 1). The exclusion of deviant points is reasonable if we consider that we do not expect temperature anomalies and suicide residuals to be coupled fully, unlike a univocal cause-effect relationship: it is self-evident that a human variable as suicide occurrence is linked to several causes that include also a simple stochastic occurrence, which may deviate from the general trend due to an underlying main variable. The detection of a deviant point is therefore a useful exercise to take these possible stochastic non-causal occurrences into account in a critical way. The exclusion of deviate points made even more evident that the links between anomalies of temperature and suicides concern the warmest months…
As pure a stream of unmitigated bullshit I have rarely seen short of a communist party broadside. There is nothing in this paragraph that makes the slightest sense; each time I read it I am exasperated. Note carefully how they try to avoid admitting the fix by using the words “our measure of temperature and suicide” which here literally means “the two series we created by several different methods but which are not temperature and suicide.”
The whole point of this gibberish is to say, “We could not find a signal before we manipulated our data, nor could we find it after we manipulated it twice successively. But were were finally able to produce what we were looking for by throwing out the points that did not fit our needs.”
Only after they tossed the data (and after the manipulations) were they able to get publishable p-values. And not too many either: they only found a relationship in a couple of months and really only in males. As cheaters they stink. But not entirely. They did (12 + 4)*2 = 32 separate tests (one for each month and season once per sex), and they should have adjusted their p-values up to account for the multiple chances for success. They did not. Well, no time be ethical now. If they did adjust, their findings would vanish.
The only question I have is: did they cheat knowingly or were they so anxious to justify their fears of global warming that they actually believed that anything they could do to the data would be a service to humanity? Sigh. Probably the later.
The comedy really begins when the discuss their findings. Our authors have learned to write badly as all academics must if they want to see their names in print. Try this fragment
At a first glance, more suicides in the months with a greater thermometric discomfort can be attributed, at least partially, to the effect of global warming on survival after attempt. Very simply, extreme temperature might increase the negative impact of body self-harm on the chance of surviving after impact…
“Thermometric discomfort”? “Negative impact of body self-harm”? Oh, Lord. At least they later admit that “No doubt that thermometric discomfort can cause stress even when associated with increasingly cold temperatures.” By which they mean, cold weather can be awful too. Fewer suicides might—might—be found in winter because people are stuck inside and “are more likely to get an enhanced surveillance by their families.”
The authors are willing to grasp at any reason, any connection no matter how tenuous, that shows they are right. Thus,
As correctly pointed out by an anonymous reviewer, the same mechanisms affecting climate (anthropogenically-induced multi-pollution) are also affecting human health, as seen in the increased incidence of cancer in younger people and in the earlier onset of neurological disease in adults in the Western world…This may, as well, increase the risk of suicides, which is itself enhanced in these illnesses…leading to a spurious correlation with anomalies of temperature that are interrelated with all the other effects of multi-pollution.
“Multi-pollution”? Yeesh. Stick with me, it gets worse…
Moreover, the waste of natural landscapes and the deterioration in flora and fauna related to multi-pollutions and its effects on the natural world might further reduce the sense of satisfaction had happiness in people, excursion in natural environments representing an opportunity to buffer the negative stress of urban life.
What’s the word that describes wanting to cry and laugh simultaneously? That word is me all over after reading that.
Each sentence in this work is a gold mine of unintended comedy and shoddy work. I could write a book on How Not To Do Research based on just this paper. However, I am growing weary and have probably tested your patience far longer than I should have. So we will leave off with their overall conclusion:
An improvement in the ability of communities to adjust to temperature changes by implementing public health interventions may play an important part in preserving the wellness of the general population, and also in limiting the worst consequences of suicidal behavior.
” Worst consequences of suicidal behavior”? Oh, I give up.