I applaud the fact that they admit to the mistakes and owning up to them. Converting abstract relationships into semiological representation is not exactly well cut science. There is a bit of experience and discovery in being able to take complex data and present it in a clear manner.
Now, only if we can get publications like Bloomberg to admit their mistake about "The Big Hack" but off-topic and I digress. I've always enjoyed Economist and its British humor, and this article just adds more credibility.
The Economist’s willingness to fess up to their mistakes makes me trust them. For years, even up till 2017 every mention of the Iraq war would include the caveat “(which this newspaper supported)”. They didn’t need to, especially since other publications don’t do the same. It shows a level of integrity and self-awareness that appears generally lacking.
For me their support for the Iraq war was where we parted ways permanently. Anyone with half a brain could see what was going on in the build up to Iraq. The Economist was for _intelligent_ analysis.
They could have chosen to be honest “The Iraq war is a fantastic opportunity to plunder the wealth of a nation, therefor as Economists we support it”. Instead they got onboard with selling a war based on false premises.
The minimum would have been simply “we are choosing not to cover the Iraq war as we feel it puts us in a position of being intellectually dishonest”. Or they could have taken a bolder stance and questioned things like what it means to declare war on a noun (war on terror) instead of a nation.
Ultimately, considering the number of people that died due to the Iraq war and the ensuing chaos, simply putting “(which this newspaper supported)” in brackets seems like a lame response to me. Instead how about retroactively analyzing what really happened, what mistakes the Economist made and why, and publishing that outside of the paywall?
>They could have chosen to be honest “The Iraq war is a fantastic opportunity to plunder the wealth of a nation, therefor as Economists we support it”. Instead they got onboard with selling a war based on false premises.
You're suggesting they knew the premises were false at the time. They very likely did not know that at the time; nor did any of the major media outlets.
Yes, they all, including the Economist, could've given the war much more scrutiny and skepticism from the start, but that's why they're admitting they messed up every time they mention the Iraq war.
You also have to consider the politics and atmosphere at the time: most people in the country (regular citizens, politicians, newspapers, everyone) supported the war and truly believed that Saddam Hussein had weapons of mass destruction. We all already knew he was a brutal despot who had no issue using terrorism and WMDs against civilians in the past to achieve his goals, so it's not like his having WMDs in 2003 would've been the most shocking revelation in the world. It just happened to be untrue.
This is a massive collective mistake, not just something the media screwed up. The media certainly played a big part, and the media definitely needs to be held to higher standards than regular citizens do, but the blame can't be placed solely on their shoulders.
>Instead how about retroactively analyzing what really happened, what mistakes the Economist made and why, and publishing that outside of the paywall?
I agree that's a great idea. I think they should do both (a full retrospective + disclosing they supported the war when they mention it in the future). They could even hyperlink "(which this newspaper supported)" to point to the analysis.
> You're suggesting they knew the premises were false at the time. They very likely did not know that at the time; nor did any of the major media outlets.
Everyone paying even little attention to anything other than US propaganda knew it, among other times, when the US “evidence” was debunked by UN inspectors in the same hearing it was presented. Which the major broadcast media all transmitted live, so if they were watching their own coverage they knew it, too.
> We all already knew he was a brutal despot who had no issue using terrorism and WMDs against civilians in the past to achieve his goals
We (those of us who had been paying any attention, at any rate) also knew that much of the senior national security leadership of the US administration that was pretending to be concerned about that had, in fact, actively, vehemently, and overtly supported those actions at the time, and that their later pretense of concern about his past behavior was as fabricated as their claims about his supposed present behavior.
>Everyone paying even little attention to anything other than US propaganda knew it, among other times, when the US “evidence” was debunked by UN inspectors in the same hearing it was presented. Which the major broadcast media all transmitted live, so if they were watching their own coverage they knew it, too.
I think that's a convenient story that sounds great in hindsight, but I don't see evidence that that's what actually happened at the time. Yes, the Chief Inspector tried to dispute some of Powell's claims, but this was not necessarily considered a "debunking".
There was a great deal of confusion, uncertainty, and disagreement, including among the UN inspectors themselves. It wasn't black and white. You can't point to any sort of convincing proof, pre-invasion, that Iraq did not have WMDs.
Of course, one shouldn't have to prove a negative, and one should have very damning proof if they're going to start a war, but the point is that a lot of smart people did think there was a legitimate risk. They were just wrong, and in some cases were misled.
The media should have been much more balanced and cautious with their reporting, for sure, but the suggestion that all the media outlets just intentionally lied to everyone about WMDs is very unrealistic. There certainly may have been lying going on inside the intelligence community and perhaps the executive branch, but that's not the same.
I do think a big takeaway from the Iraq War and some of the Trump stuff is that the mainstream and historically respected media (NYT etc.) are much less trustworthy than we all initially thought. But they're less trustworthy because they're fallible humans who often succumb to biases and overconfidence in their analyses - not because they're deceitful evildoers.
>We (those of us who had been paying any attention, at any rate) also knew that much of the senior national security leadership of the US administration that was pretending to be concerned about that had, in fact, actively, vehemently, and overtly supported those actions at the time
> You're suggesting they knew the premises were false at the time. They very likely did not know that at the time; nor did any of the major media outlets.
There I have to disagree. Strongly. At the outset of of the Iraq war it was already clear to anyone working in journalism that Reuters and AP Newswire had an unhealthy grip on the news industry. As an outsider it was clear simply by observing that _every_ news source was beating the same drum and pushing the same talking points.
I remember pretty well the build up - my daughter was 1.5 years old and my life at the time was work, changing nappies, and listening to the news. The Economist was one of the last newspapers to chime in on Iraq, and I had hopes they’d at least take a neutral, objective position. Instead they got onboard selling the war, presumably because they were afraid of the consequences of failing to do so - remember Bush declared “you’re either with us or against us”.
The Economist actually played a key role ... basically the final nail in the coffin in bringing the intelligentsia onboard with way where only Germany (Schröder) stood against.
Anyway that’s all in the past. But I won’t be subscribing to the Economist any time soon
>You're suggesting they knew the premises were false at the time. They very likely did not know that at the time; nor did any of the major media outlets.
It was absurdly obvious.
At an absolute minimum for balanced reporting, they should have clearly reported both the official "WMD" line and the fact that a massive number of people were calling that assessment bullshit from day 1.
Exactly, they are by far the less biased news outlet I read. For instance, they explicitly don't support Trump, but don't hesitate to say "Trump is right about X", or "Trump did well on Y". They also don't support the Chinese PC, but don't hesitate on giving recommendations that would help them fix their problems (knowing that the Chinese elite read The Economist)
Saying they aren’t biased doesn’t mean they don’t have a viewpoint, which they themselves disclose:
“We do not attach ourselves to any political party. Our public agenda is liberal in the classical sense. We have supported free trade ever since our foundation in 1843 when we opposed Britain’s corn laws, which sought to keep the price of grain high by limiting imports. We have continued to advocate bold policies in favour of individual freedoms, such as same-sex marriage and legalisation of drugs, regardless of whether they are politically popular, in the belief that the force of argument will eventually prevail.”
We (Norway) cheat a bit, though - we're a small (5M and change citizens) economy with a massive oil extraction industry which we tax to high heaven, hence the large fraction of GDP raised by taxes.
Regular corporate tax is 22%, whereas anything involving pumping hydrocarbons out of the ground is taxed with an additional 56%, leading to a tax rate of 78%.
In 2018, this led to more than $13B in taxes being paid from oil companies. Joe Q. Public, for comparison, paid $35B or so in income tax.
In my defense, I hardly slept last night courtesy of a three-year-old with the chicken pox, and I hadn't had my morning gallon of coffee when I replied.
Personally, my tinfoil hat theory is that it was concocted by the white house staff to create a political climate conducive to playing hardball the Chinese. Not one shred of evidence, of course, but it seems the most plausible thing I could think of.
The most surprising conclusion was about the final plot discussed, regarding scientific publications. The author says that it contains too much information and is confusing for that reason, and postulates it might be better to summarize the different fields and plot as a single variable.
The truth is, it is not a confusing visualization. There is no rule that says to remove variables for the sake of reducing a variable count. It is useful to see that health sciences stands apart from other fields, and other points can be seen as a cluster or examined more deeply by the reader. It is actually one of the best plots shown in the entire article.
In effect it is a grouped bar chart, a visualization that many publications shy away from in favor of stacked bars. While that is good for showing a total, it is often a less effective way of visualizing data—it's difficult to compare magnitude of anything but the base bar. That fault can be seen in the stacked bars of the budget surplus example, which still exists for the improved version.
At any rate, I'm glad the Economist is looking to improve their visualizations and isn't afraid to share mistakes.
On the dog weight and trade deficit charts, I'd argue there's merit to the originals. The essence of a linear relationship between two data series X and Y is that X and Y are identical up to position and scale. The "better" charts represent the data more accurately, but the original charts make it easier to see the relationship between the two series.
Also the trade deficit chart is confusing in part because the negative sign isn't obviously a negative sign.
Edit: the last chart is fine. It's maybe not very sexy, and it takes a moment to absorb, but the data is clearly displayed.
The dog weight one is a good case of how you can futz with numbers to imply a greater relationship than is present. I think the second chart is much better in that respect. If you're doubling the relative range of one of a double scale to make stuff look like it fits, that's not exactly being forthright.
> but the original charts make it easier to see the relationship between the two series.
Well, as the article says, if it looks like they're a perfect fit when they definitely aren't, that's probably beyond making something clearer and going into convincing people of something that the chart doesn't really imply.
> the last chart is fine. It's maybe not very sexy, and it takes a moment to absorb, but the data is clearly displayed.
Totally agree on this. It takes a bit to digest, but that's because there's a lot of information imparted.
There's no reason the scale needs to be the same when comparing measurements in totally different dimensions.
The weight is roughly a volume measurement while the neck size is a linear one. Also, if the necks were measured in inches, you'd end up at roughly the original chart, even being mindful of the scaling on each axis.
I think the left chart is actually the better one. It does a good job showing the correlation between neck size and weight.
>The dog weight one is a good case of how you can futz with numbers to imply a greater relationship
What they did was very similar to standardizing data series (normalizing to mean=0 and standard deviation=1). This is a standard tool in scientific data analysis (most published papers use it) and it's taught in statistic and data science classes. It's also a preliminary step in computing most correlation measures.
It does not exagerate relationships, it helps visualizing them.
The second chart removed the standardization and obfuscated the relationship.
As you can clearly see from the axis labels, they are not plotting standardised or indexed data, what they did was hand picked a rescaling that makes the plots line up very nicely.
Further, it is simply not true that weight is a perfect function of neck size, so you shouldn't make a plot that implies that.
>what they did was hand picked a rescaling that makes the plots line up very nicely
I bet dollars to donuts that's what a stardization would show too, unless I'm missing something. It would show an even more perfect lining up because it would line up the line ends.
Wish I could test it. Wish pandas had a read_image function to get data from line plots like it has read_html.
The reason it's hard to pick which chart is better is that both charts are far too detailed for the tiny amount of information they convey, which is "both values shrank over time"
Can you further explain the frankly astonishing statement that you find it hard to pick between a plot showing a false equivalence and one that doesn't?
As you suggest, it's not always incorrect to shift and rescale data for visualisation. In fact, that is exactly what a linear correlation calculation does: the data is shifted to have mean zero and rescaled to have variance one, then the covariance is computed.
The real problem is that often, the two timeseries being compared are too simple to demonstrate whether there is a relationship present. If they're both essentially straight lines as in the Butcher's Dog graphs, you may see a spurious relationship.
What you need -- in unscientific terms -- is to see one timeseries do something weird, and the other timeseries reacting to the weird thing in a manner that fits the narrative. Straight lines don't really do that, but often the first difference of two timeseries that look like a straight line will be good enough.
I don't get the dog one at all. The article is about dogs getting smaller. Since neck size and weight are very correlated why bother to show both? I would just have a single plot showing the decline in dog weight over time. The fact that smaller dogs also have smaller necks is neither surprising nor useful.
My favorite mistake The Economist made was a survey they ran. They compared UK to US, and supposedly one of the questions was:
What better explains the origins of the Earth?
A) Evolution.
B) The Bible.
It's unclear whether they literally asked that question, or they asked about origins of life and just had a copyediting mistake in the summary of the results. They ignored my letter about it, though, so I suspect the former :)
Yeah, evolution didn't get started until after the Earth had already originated. Not saying that "In the beginning God created the heavens and the earth" is the best explanation, but it's certainly better than what evolution gives, which is "after the earth already existed..."
This thread is assuming biological evolution, but dynamic systems in general evolve. Maybe I'm stretching the word too much, but the question was a trick question to begin with.
I mean... The dictionary definition of "evolve" is basically "things happened over time". But especially increasing in complexity, although it's not a necessary qualification.
So basically, as long as there is any temporal component, it "evolved". But I'd argue that saying it happened doesn't explain how it happened. So the "correct" answer is still B.
Is your statement 'Saying that it evolved is not saying how it evolved.'?
We have planetary and solar system dynamics fairly well understood in terms of models and matching that with observational data. So I think an astrophysicist could make the how argument from an evolutionary perspective.
Your initial point was that we should not assume "Evolution" means biological evolution. Let's remember that and proceed:
In a pedantic sense, saying that "it evolved" means nothing because everything "evolved" for some suitable non-Darwinian interpretation of what "evolved" means. For example, if you're willing to not assume "Evolution" implies "biological evolution" then why not argue that woman evolved from man's rib as per the Bible and that man "evolved" from dirt?
"Evolution" without explicit context, and perhaps additional qualification (e.g. "the evolution of our solar system") means that as much as it means nothing.
I think “I don't know, that's not my department” is a better explanation than an elaborate one which purports to address the question but which is both inconsistent with itself and, whichever way you resolve the internal inconsistency, also with the external evidence, so even reading “evolution” as “the theory of evolution by natural selection”, I think it's the better explanation.
But “better” always involves a subjective quality judgement.
Their point is that evolution doesn't explain the origin of Earth, a planet, at all. While the Bible's explanation may be fiction, it does offer an explanation for the origin of Earth, and is therefor the better explanation by virtue of providing an explanation.
No, no, it does, you see. Everything evolves! Everything from the lowliest quark to the grandest Londoner desires to become greater than it is, to achieve the pinnacle of being, to immanentize the eschaton, by becoming a phony-doctor of computer science from Texas.
It is a family tradition in my house to pore through the Economist every week looking for the worst visualization. So, props to them for admitting they have a problem.
I do the same with CBC. The offense I hate the most is when they chop off a bar chart so one bar is twice the height of another but the actual value is only like 5% higher.
Wouldn't you find more in other newspapers? I generally find Economist visualizations surprisingly well made. Articles like this show that they're actively trying to improve.
I've noticed the scales issue happens a lot with claims about markets. A story will say something like 'Asset A plummets among bad event X' with a daily graph of the price. When you look back at the weekly or monthly graph, though, you find that it's well within normal fluctuations and isn't even a low for that period.
I'm an Economist subscriber and so read it most weeks, but honestly their charts and visualisations are often terrible. I think they could read a bit of Tufte and reduce chart junk as a starting point. This article does make some other good improvements though, so there is hope.
My college statistics textbook had an entire problem set where every question was dedicated to a bad or misleading chart from "USA Today". These are mild in comparison, but it's important to get visualizations correct, given how powerful they are at communicating ideas and concepts.
This is a confusing graph from the Economist that I snipped when I saw it in 2013. Very difficult to wrap your head around what is being shown here: https://pasteboard.co/I7ttkG5.png
I actually think that this might be one of those cases where the borders of a graph matter a lot. The way that they styled it here makes it look like a time series, and makes it really difficult to interpret at first sight. I think flipping the axes to make it much akin to a cumulative distribution function would make it easier to read.
I assume you wrapped your head around it by now, but I'll explain it anyway.
Bottom right; barely 5% of the world population is equal or richer than the average US citizen. 20% of the world population has at least 25% of the average US citizen, or 80% of the world has less than than 25% of the average US citizen.
Note that a graph of the US alone would look about the same.
I think its a highly informative graph, that you will not get at a glance. Another mistake they noted in the article:
"Another odd thing is the choice of colour. In an attempt to emulate Labour’s colour scheme, we used three shades of orange/red to distinguish between Jeremy Corbyn, other MPs and parties/groups. We don’t explain this. While the logic behind the colours might be obvious to a lot of readers, it perhaps makes little sense for those less familiar with British politics."
I think it's a bonus if you can make a graph have extra layers of information for the informed reader. In literature this his highly regarded. Why not in data visualization?
> Bottom right; barely 5% of the world population is equal or richer than the average US citizen.
But GDP per capita is not a measure of personal income or wealth, it's a measure of production of a country. Since it not a measure that can be applied to people individually, it seems ambiguous what it means for a given person in the world to be "above a given % of US GDP per person."
US GDP per capita in 2017 is about $60k. What does it mean for an individual in the world to be "above" this? Their personal income in a year? Their total wealth?
My best guess is that this graph considers a person "above" this if they live in a country that has a per capita GDP above $60k (at purchasing-power parity). So really it's a comparison of countries and their per capita GDP, weighted by population.
So I think the bottom right is actually saying "5% of the world population lives in a country that produces more per person than the US." It doesn't say much of anything about how rich individuals are.
The disparity between the (very) rich few and the many poor would result in the share of people above and below the hypothetical "average" person not being 50/50.
Perhaps it would be more clear sticking to a single value of US GDP, e.g. as "% of world pop. above 1/4 of US GDP per person, over time", but the data shown only has three time points ...
If you can show me the specific error, I'd be interested to look into this, since I work on Let's Encrypt, which provided the certificate for that site (and it's useful to know about browser compatibility problems or new potential kinds of misconfiguration).
An error occurred during a connection to pasteboard.co. Peer received a valid certificate, but access was denied. Error code: SSL_ERROR_ACCESS_DENIED_ALERT
The page you are trying to view cannot be shown because the authenticity of the received data could not be verified.
Please contact the website owners to inform them of this problem.
I enjoyed the post. I am not familiar with the website, but from the examples in the post, it seems that only the x-axis is labeled. It can be hard to understand the meaning of a plot when only one axis is labeled.
I'm currently reading Statistics for Dummies, and these kind of techniques used i.e. wrong scale, incorrect starting point, etc is explained very well. Highly recommended.
Now, only if we can get publications like Bloomberg to admit their mistake about "The Big Hack" but off-topic and I digress. I've always enjoyed Economist and its British humor, and this article just adds more credibility.