[Editor’s note: You’ve heard there are “lies, damned lies, and statistics”? That it’s easy to lie with statistics—even if your numbers are true? Those aphorisms are absolutely true. And what our friends at the CO2 Coalition reveal in this brief article is a stellar example. Lancet, which is among the most respected medical journals, should be ashamed to have published the incredibly deceptive graph shown on the left. Thanks to the CO2 Coalition, the graph on the right corrects it. Climate catastrophists are notorious for such statistical sleights of hand, but this one deserves particular condemnation. Note well: See the correction of this editor’s note below the article.—ECB]
It is very well documented that many more people die from cold than from heat. The largest study on deaths attributable to heat or cold found that cold weather kills 20 times as many people than heat. Another study in the U.K. and Australia found that cold-related deaths in these countries accounted for more than 15 times higher mortality than heat.
The results of a new European study in The Lancet reported that cold-related deaths account for 10 times the number as deaths due to heat. But when it came to presenting the results, they pulled a graphical sleight of hand.
Chart “A” on the left is from The Lancet article. Note that the X axis at the bottom uses two different scales for heat and cold mortality. There is a five-fold difference that accentuates heat-related deaths and minimizes deaths due to cold.
Chart “B” on the right was created by the CO2 Coalition technical staff and accurately depicts the huge disparity.
Nice try, but The Lancet has just been exposed. Warmer weather would likely save millions of lives.
This piece was originally published by Co2Coalition and has been reproduced here with permission.
[Correction: The editor’s note and the article above attributed deceptive intent to the authors of the Lancet article. A reader protested in the comment below, to which we replied in another comment below. The reader then wrote us again, arguing: (1) “Reading the ‘incriminated’ Figure 3 of the Lancet article in context, I would never have imagined the suggested deceptive intention on the basis of its format.” (2) “In its summary, under Findings, the article clearly says: 203 620 of annual excess deaths attributed to cold and 20 173 to heat.” (3) “… on page e277, just before the discussion of Figure 3 these numbers are repeated.” (4) “… the figure caption and the marked text passage explain the purpose of Figure 3. This purpose is to show raw death excess rates broken down by age groups at the country level.” (5) “In my opinion, Figure 3 has nothing to do with the attribution ratio of heat versus cold excess deaths, nor with suppressing information about this topic.” (6) “The alternative presentation that you (and the CO2-Coalition) do propose, suppresses the information intended by the authors, both by suboptimal scaling and by deleting the coding of colors for the age groups.” He concluded: “If I had been one of the authors, I would be offended by your unfounded accusation of a hidden motive. If you want to maintain that accusation, it is so. I find it an attack of integrity without any substantiation, which may be ignored by any researcher.”
We appreciate Mr. Verjheij’s care in arguing his case, we are persuaded that we were wrong to attribute intentional deception to the article’s authors and to The Lancet, and we apologize for doing so. However, we still firmly criticize the graph as misleading—even if only unintentionally so, for three reasons.
Here is the Figure 3 as published in The Lancet:
First, the age-range information could have been communicated effectively while putting both heat and cold on the same horizontal scale. That being so, one must wonder, “Why, then, did the authors choose to use such different scales—a 5-fold difference, in fact—for heat and cold?” Can they seriously not have recognized that doing so would give pretty much every viewer, at least on initial view, the impression that the ratio of heat-death rates to cold-death rates is about 5 times higher than it actually is? And can anyone who understands human psychology, summed up in the aphorism, “You never get a second chance to make a first impression,” not have understood that it would be very unlikely that that false impression would be reversed unless people carefully read the specific bits quoted and noticed how they differed from the impression given by the graph? Consider the fact that most people, even subscribers, don’t actually read the full text of articles in technical, refereed journals. Headlines, abstracts, opening paragraph—those are about as far as most readers go.
Second, there is another aphorism, “A picture is worth a thousand words.” If we know anything about human visual (text and pictures) communication, it is that pictures grab much more attention than do words. (I mourn this. John 1:1 and the fact that God didn’t include any sketches, paintings, or other forms of visual art in the Scriptures imply it should not be so.) Anyone who understands visual communication would know that the graph would make a strong impact on readers/viewers long before they reached the verbal explanation.
Third, communication is not simply what the speaker/writer intends but also what the hearer/reader understands. We will concede that those who actually took the time to read the text (probably a small minority of those who, thumbing through the issue of The Lancet and picking up quick ideas—for very few do read the full text of such articles), and who also thought about how what they read should inform how they would interpret that graph, should not have been led astray, and probably would not have been. We will also assume, now, that the authors did not intend to mislead regarding the ratio of cold to heat deaths. Nonetheless, putting cold and heat deaths on such radically different scales, particularly when each scale ended with “250” (lower left and lower right), was bound, despite the easily overlooked discontinuity bars on the X-axis just before the “250” at lower right, to create an impression as to their ratio that differed radically from that in the actual numbers, and the authors should have known that. One of the most important tasks of a communicator is to anticipate how an audience is likely to misunderstand and then preclude that. In this graph, these authors did not achieve that.
So, we retract the attribution of intent to mislead on the part of the authors. But we sustain our critique that the graph is, in fact, misleading to all but those who carefully read the accompanying text and think about how it should change their natural interpretation of the graph.—E. Calvin Beisner]
E. Calvin Beisner says
A reader wrote this to us via email, as a Letter to the Editor:
Dear editor,
I think that the message of this CO2 article and your reproduction is unfair and uncritical.
1. The insertions on the left-hand side of the figure, respectively 1-250 and 0-50, are not in the original Lancet-article, but added by the CO2 Coalition.
2. Obviously to me, the objective of the properly indicated different scales on the cold and heat related excess death rates, is different from the one insinuated by your reproduced article.
The Lancet figure shows, in addition to total numbers, distribution over age groups (see also the information on the top of the figure and in the article).
This information is lost, respectively, deleted in the presentation proposed by the CO2 Coalition.
In my opinion, the nature and the full content of the scientific article justify the efficient way of presentation applied by Lancet The change and deletion made by the CO2 Coalition are more manipulations for their own ‘advocacy purposes’ than science based improvements.
Best wishes,
Jan W. Verheij, The Netherlands
**********
Our response follows:
Thank you for sharing your concern with us. The CO2 Coalition’s complaint, which we share, is that the graphic representation of the deaths from cold versus heat, regardless of age group and regardless of country or region of occurrence, is highly misleading (to the point we would call deceptive) in the study published in The Lancet. Here’s the original graph, minus the CO2 Coalition’s revisions:
[We then inserted The Lancet’s original graph.]
The clear impression is that heat deaths are only marginally fewer than cold deaths. What the actual numbers show is that, again regardless of age group or country/region of occurrence, cold deaths exceed heat deaths not by a little bit but tenfold (as the actual numbers in that study showed). There’s no reason why that study couldn’t have put the cold and heat deaths on the same scale—or, rather, the only reason we can figure is that it wanted to give the casual reader (and the general Lancet reader, and those who see the graph as picked up by other sources, are unlikely to delve into the actual numbers) the impression that the difference between cold and heat deaths is marginal.
That the CO2 Coalition’s redrawing of the graph loses age groups is correct, but their complaint wasn’t about age groups but about the misperception the study’s original graph created by using the different scales for heat and cold.
We stand by the critique.