To think this qualifies as bad science?

[tamade]

The graphic below is from a Lancet article about the relative impact of cold or heat on excess death rates. Look at the scale(s) at the bottom and tell me that they were not deliberately chosen to mislead the casual observer.....

YABU climate is so important the ends justify the means
YANBU it is a bit dodge and all scientists should hold themselves to better standards
bonus YABVU its obviously an honest mistake and it makes the graph look better: stfu

http//doi.org/10.1016/S2542-5196(23)00023-2

If you aren't 'meant' to be comparing the red and blue sides, they could be placed in separate graphs. By producing the graph like that it invites comparison.

It may be the intended audience are OK with it, but once a graph is 'out' in public it can be misleading. You can go away with an 'impression' of the data that is entirely incorrect.

People die a lot more from being cold?

I'm pretty stupid, but could see the sets of different scaled numbers on the bottom axis of the graph. Why is it misleading?

It wasn't on a public press release though was it? Context is important, even if it's not the clearest graph / example of best practice I'd expect Lancet readers to be capable of checking the axes especially when the figure is referred to in the text. I don't understand why you've taken this so seriously.

As soon as people notice they will think that they are being fooled therfore they won't take it seriously like they should

Poor practice

It’s very bad practice to change scales on a graph like that. And they’ve pointlessly added the //250 at the right hand side of the heat axis. Would have been better to place two slightly separated graphs side by side if they really wanted separate scale bars to indicate the detail on the hot scale.
This is the problem with bad data science representation like this. It’s not lying, but it’s initially misleading, and you show graphs to visually indicate patterns and relationships as well as to represent the data.

Good data viz makes it easy and desirable to interact with the data represented. This is not good data viz.

Got to love a misleading scale - normally seen in the media when they want to mislead. For example, don't start at 0 so the differences seem huge.

I suppose that if they had used the same scale on both sides, then it would have been hard to see the differences between countries for the heat as they are quite similar.

2 graphs would have been better.

Not read the paper - but I guess it does draw some conclusions about the effect of heat and cold.

It's even more confusing with the age groups in there as well.

Too much going on

It’s like this one - yes, the scale is accurate but the graph is very misleading (and funny..)

I think the typical reader of The Lancet is well capable of reading a graph accurately

This is a subscription journal for doctors, not something carried in newsagents

And there's a mistake - <85 should be >85 (pretty sure it should actually be >=85 but this age group is reported inconsistently throughout so who knows)

I agree it's a distorted graph and purposefully so. Anyone watched Dopesick, about OxyContin? Purdue Pharma used a distorted graph to underplay how addictive it is, and look what happened there.

It's not the best presentation, but it's still fairly clear that the intention is to compare different places' rates/distribution of excess heat related deaths with other places' excess heat related deaths and then cold with cold.

I can cope with the different scales on different sides (and cold still "looks" more dangerous than heat even with the distorted scale) but the use of colours is hard to follow.

Is there a typo so that the darkest shade should mean >85? Or is the darkest shade a total up to age 85? There are five shades on the key but each row only seems to use four. Which four? Why are the "total" colours darker than the other colours? Maybe the paper explains it all but it's not easy to follow.

Looks like trying to cram too much information into a single graph.

I think you are looking at the graph first and thinking, what are my first impressions of this graph, whereas it’s creators were looking at the data first, and thinking, what is the best way of simply communicating this data. I cannot think of a simpler way they could do this, and I agree with previous posters that if a consistent scale were used, the detail within the data would be lost. If two separate graphs were used, the ability to compare the data easily would be lost.

There are 5 age groups reported but you have to enlarge the figure to be able to differentiate between the palest blues. It's shit.

The graph is fine in an academic journal. The readership have the nounce to look at the details behind the visuals. I have only seen the graph rather than read the article but to me I don’t think the aim is to be misleading, just the graph shows the trends the data suggests.

if the article was in a newspaper and included for sensationalism I might think differently.

It's quite a widespread thing actually, to use accurate data but with misleading visual interpretation to push some kind of agenda. Yes, a lot of people will actually read the scales and understand/pay attention to everything, but a lot more people will just remember the imagery.

This is starting to be covered more extensively in y6 and it's always fascinating to watch the kids get that lightbulb moment, as is their outrage.

It's been doing the rounds on LinkedIn, mainly shared by various groups / people with an agenda to down play climate change from what I've seen so far.

Oddly enough I note the "issue" is the perceived exaggeration of heat related excess deaths, yet none of these so called climate and data experts, who have an issue with the graph, are picking up on the fact that:

a) the graph relates solely to European countries which generally have a colder climate than Equatorial and Southern hemisphere nations and so you would naturally expect there to be more cold related deaths, so we aren't making a global comparison to a global problem.

and

b) the author's haven't allowed / adjusted/ corrected for non-weather related deaths such as the seasonal cycle of excess deaths caused by flu, which, again in Europe, tends to be significantly higher in winter than in summer.

It's almost as if the graph is aimed at people who understand the topic and not for the general public who can't even tell when a massive red bus is covered in a blatant lie.

I am a Ph-D level health scientist and nope, I don't take any issue with this graph. If I understand it, you're having a problem with the fact that the authors have make the range of the excess deaths the same for heat/cold attributed deaths, but there is higher excess mortality for cold than hot, but at first glance, it looks similar ish in magnitude because of the right hand side of the scale as been truncated to fit.

This really isn't a problem at all - and to be fair, it's quite likely they submitted the graph with the scale the same on both sides to the Lancet and then got asked to change it because it didn't meet the formatting guidance for how they can publish figures (it can be a massive pain!). I don't have any issue interpreting that graph though and I don't think any other scientist would either. Now the Daily Fail may freak out and run with something click baity because their journos don't often know how to interpret scientific findings, but I don't think there is anything scientifically misleading here.

@mindutopia
I too am a ‘PhD level scientist’ though not in health; I run a laboratory which is responsible for releasing materials for use in aerospace and expert witness work in materials failures. It’s probably a different kind of science, like American and English maybe