This chart is a work of art

June 25, 2021 5:08 AM

Katelyn Gadd highlights a CNN chart which is more than a little misleading. Bonus: more terrible charts!
posted by Stark (46 comments total)

This is less "ha-ha" funny and more "manufacturing consent for the police state" horrifying.
posted by Glegrinof the Pig-Man at 5:16 AM on June 25, 2021 [43 favorites]

What is it even a misleadingly presented chart of, purportedly? It doesn't have a sufficiently detailed caption.
posted by eviemath at 5:19 AM on June 25, 2021

I believe it's supposed to be the % of adults who report that they think violent crime is a problem. I'm sure media reports have no influence on this statistic
posted by dis_integration at 5:26 AM on June 25, 2021 [9 favorites]

Still more terrible charts (with an excellent URL): WTF Visualizations
posted by adamrice at 5:44 AM on June 25, 2021 [5 favorites]

Same wine, upside down bottles.
posted by bonehead at 5:44 AM on June 25, 2021 [5 favorites]

Also the % of adults who experience time backwards.
posted by adept256 at 5:49 AM on June 25, 2021 [15 favorites]

As far as I can see there are a number of problems, and I'm not an expert in visualisations:

1. The y-axis doesn't start at zero
2. The y-axis isn't labelled correctly (it should be % of Adults who report that they think violent crime is a problem)
3. The gaps between the data points on the x-axis are not representative of the gaps between the dates
4. The margin of error is not highlighted on the chart's data points
5. Straight lines between the data points imply a steady trend which may or may not be the case
6. In my opinion the most amazing weirdness is that the x-axis dates go from right to left!

posted by Stark at 5:57 AM on June 25, 2021 [10 favorites]

The design is Ed Wood level awful.

And a reminder that US tv news loves, loves, loves to stoke fear of crime. CNN et al are major contributors to the American Gothic.
posted by doctornemo at 5:59 AM on June 25, 2021 [10 favorites]

Just amazingly blatant manipulation. Nothing about that chart makes sense, except as pure indoctrination. Tells you all you need to know about mainstream media complicity in the carceral state. Mefite Andy Baio shows what the chart would look like without the obscene distortions in one of the comments in that thread.
posted by mediareport at 6:00 AM on June 25, 2021 [21 favorites]

I dunno, this feels more like gross incompetence than agenda setting. I base that opinion partially on the fact that I couldn't figure out what the chart was trying to show.
posted by gwint at 6:01 AM on June 25, 2021 [1 favorite]

On the other hand it does seem you'd have to work particularly hard to make a chart this bad. Like really solid effort.
posted by gwint at 6:04 AM on June 25, 2021 [11 favorites]

Those were clearly intentional decisions, to me, anyway.
posted by mediareport at 6:06 AM on June 25, 2021 [16 favorites]

The selected months are also a problem. The months are not equally spaced and they choose June 2020 as the nadir of concern over crime. Probably the maximum month of lockdown.

And finally the question is stupid. Of course violent crime is a concern. And baby bunnies are fluffy.
posted by dances_with_sneetches at 6:51 AM on June 25, 2021 [7 favorites]

I just don't think you can make a backwards chart accidentally. That requires willful intent.
posted by graventy at 6:55 AM on June 25, 2021 [18 favorites]

graventy: "I just don't think you can make a backwards chart accidentally. "

No, someone would have to be pointing a gun at your head. And 51% of adults think that pointing a gun is a crime.
posted by chavenet at 6:59 AM on June 25, 2021 [6 favorites]

I just don't think you can make a backwards chart accidentally. That requires willful intent.

You could actually end up with this chart accidentally if you put the four data points into Excel and used the default sort function. Note that the labels on the x-axis are sorted alphabetically.
posted by Tau Wedel at 6:59 AM on June 25, 2021 [43 favorites]

I have an incoming batch of summer analysts who are taught explicitly about misrepresentation of data and how poorly things can go for you when you do that. I noticed the alphabetical sort (not actually the first time I’ve seen that!!!) and this is sort of a perfect example of why you need basic graphic standards and understand what you are showing.

I’ll be using this as an example today. Thanks!
posted by larthegreat at 7:09 AM on June 25, 2021 [10 favorites]

*sigh* Does no one study Tufte anymore?

That said, I'm going to assume (pray?) this was done by some unpaid production assistant armed with their employer's monthly Adobe Creative Suite tithe. Or, has graphic design training actually devolved this horrendously?
posted by Thorzdad at 7:19 AM on June 25, 2021 [3 favorites]

Running time backwards is a classic trick; here's a graph that not only puts time backwards but puts it on the Y axis in order to flip an upward trend upside down to make Trump look good.

Tufte schmufte, that Trump chart and the one considered here are so terrible they aren't even "How to Lie with Statistics" bad. The techniques in that book are for people who are still trying to make a plausible chart. These charts are what happens when a numerically illiterate graphic designer is given 4 data points in an email and said "go make a graphic that supports our story that crime is a serious problem". They are putting on the disguise of scientific presentation but are in fact basically freehanding a drawing. Often literally freehanding; a popular technique is numbers that are wildly out of scale.

One more good source for bad and intentionally misleading charts: /r/dataisugly.
posted by Nelson at 7:25 AM on June 25, 2021 [4 favorites]

Never attribute to malice that which is sufficiently explained by incompetence. I can easily imagine someone giving the data to an intern and saying "make a graph of this" and the intern just selecting the cells in Excel and clicking the Graph button.

Tuftian competence is an acquired trait. I've seen very dumb graphs from very smart people who were not trying to be misleading, including, in one case, trend data presented as a pie chart.
posted by adamrice at 7:31 AM on June 25, 2021 [3 favorites]

Never attribute to malice that which is sufficiently explained by incompetence.

but we're talking about major media outlets so
posted by tiny frying pan at 7:41 AM on June 25, 2021 [8 favorites]

Never attribute to malice that which is sufficiently explained by incompetence.

"Malice" might not be the right word, but 24/7 content demand without proper editorial oversight is certainly the result of intentional decisions somewhere up the chain of command, and they deserve criticism for its frequent failures.
posted by Think_Long at 8:05 AM on June 25, 2021 [13 favorites]

I don’t see how the reverse ordering on the x-axis contributes to an intentional deception; it just makes the plot confusing. I think Tau Wedel is right: the person who made the chart didn’t know how to order the points chronologically, so they ordered them alphabetically.
posted by mr_roboto at 8:06 AM on June 25, 2021 [1 favorite]

And a reminder that US tv news loves, loves, loves to stoke fear of crime.

Exactly, the chart isn't even showing crime statistics but people's self-reported opinions of crime.
posted by gladly at 8:23 AM on June 25, 2021 [7 favorites]

Think_Long: " 24/7 content demand without proper editorial oversight is certainly the result of intentional decisions somewhere up the chain of command, and they deserve criticism for its frequent failures."

Yeah, that's an important point.
posted by adamrice at 8:49 AM on June 25, 2021 [1 favorite]

I'm not sure how you can give CNN the benefit of the doubt here??? Like yes OK sure you can do this by accident, but you can't decide put this on the air by accident. If putting this misinformation in front of people wasn't their goal then they wouldn't have done it.
posted by bleep at 9:08 AM on June 25, 2021 [10 favorites]

Has CNN apologized for the error? I doubt it, if for no other reason than the fact that CNN (and every other major network) has been pushing the "OMG CRIME WAVE OF THE 90s IS BACK!" line for a while now, and it's not even subtle.
posted by Glegrinof the Pig-Man at 9:13 AM on June 25, 2021 [3 favorites]

This point about newsroom lack of oversight paired with innumeracy reminds me of a video by Matt Parker analyzing a particular error of simple math that propagated on social media and was repeated without question by news anchors on TV during the 2020 primary season. It's a good video and worth watching but in a nutshell, someone noted that Michael Bloomberg spent $500 million on advertising for his unsuccessful bid, and pointed out that he could have instead just given a million dollars to everyone in America from that pool and still had money left over. Which of course is complete nonsense requiring not even a moment of thought to dismiss: there are something like 350 million people in the US, so he could have given everyone one dollar and had money left over. But this claim was parroted without question by news anchors, with the comment "it really makes you think." Which of course if that were true, they wouldn't have said it. Parker has a pretty charitable take on what he thinks is happening there, based on a few experiences he's had being interviewed on news programs, which is basically that news rooms are such a high pressure, fast-paced environment that there's no time to stop and think about what you're saying. Which leads to things like this idiotic graphic getting run: someone told an intern or something that they needed a graphic, the intern had no idea what they were doing but knew they had to produce something, the editor took a quick look and had to make a decision within 15 seconds and said "sure, looks like this makes our point," and this nonsense ended up on the air.
posted by biogeo at 9:32 AM on June 25, 2021 [8 favorites]

Okay but if the higher-ups thought it was important that what they say on air is accurate, then none of this would be happening.
posted by bleep at 9:38 AM on June 25, 2021 [2 favorites]

I don’t think the data presented are inaccurate. They’re just ineptly presented to a comic degree.
posted by mr_roboto at 9:44 AM on June 25, 2021 [1 favorite]

It is inaccurate in the sense that if you have been told that you can understand the line of the graph to tell you what the trend is, the line on this graph is misrepresenting the trend.
posted by bleep at 9:46 AM on June 25, 2021 [1 favorite]

I don’t see how the reverse ordering on the x-axis contributes to an intentional deception; it just makes the plot confusing. I think Tau Wedel is right: the person who made the chart didn’t know how to order the points chronologically, so they ordered them alphabetically.

It contributes to an intentional deception because most people who are half paying attention won't read the dates at all - they will see a trend line from left to right that is going up. Combine that with the fact that the caption suggests a rise in violent crime rather than concern about violent crime, and most viewers are certain to believe that violent crime is on the rise.

If it's not intentionally meant to be misleading, it is at least presented with reckless indifference to whether it presents the viewer with an accurate view of the data.
posted by Ben Trismegistus at 10:26 AM on June 25, 2021 [8 favorites]

"Often literally freehanding;"

There seems to be some subset of people that think "this number is bigger" is enough, and the idea that they should be proportionately representative of actual quantities doesn't occur to them.

One of the few errors this graph doesn't make, is that the y axis stays more or less consistent. I have absolutely seen graphs where someone would have evenly spaced 41, 48, 49, and 52 and acted like that's fine because the bigger numbers are higher.

I guess it does make this error along the x axis, because the number of months between each date is inconsistent, but they displayed it evenly spaced. That feels really far less a problem than putting the past on the right, though.


I also like the note that they surveyed adults. What is the trend for opinions among children?
posted by RobotHero at 10:34 AM on June 25, 2021 [3 favorites]

so I guess it's now:

1. lies
2. damned lies
3. statistics
4. charts

Thank you, Mr. Disraeli ... or whoever.
posted by philip-random at 10:55 AM on June 25, 2021

...they choose June 2020 as the nadir of concern over crime. Probably the maximum month of lockdown.

There are a lot of people who want to delegitimize the concept of defunding the police, which gained prominence in the weeks following the murder of George Floyd, by claiming it sparked an increase in crime.

Portraying June 2020 as a huge inflection point seems pretty deliberate in this light.
posted by theory at 11:33 AM on June 25, 2021 [2 favorites]

Never attribute to malice that which is sufficiently explained by incompetence

Just the other day CNN ran this piece of blatant copaganda on Jake Tapper's show:
CNN goes for ride-along with NYPD as crime wave worsens @jimsciutto reports

This is not incompetence. This is CNN's corporate agenda.
posted by Atom Eyes at 12:00 PM on June 25, 2021 [10 favorites]

2014's Misleading Gun-Death Chart Draws Fire, via LiveScience: "Recent reports about how Florida's 'stand your ground' law affected the number of deaths in the state have raised a few eyebrows." Source for chart, which "defies convention by showing higher numbers near the bottom of the vertical y-axis, confusing some viewers," is the Florida Department of Law Enforcement.

"The designer of the chart, Christine Chan, explained her decision on her Twitter feed, saying, 'I prefer to show deaths in negative terms (inverted). It's a preference really, can be shown either way.' Chan also noted that her inspiration for the chart came from a visually compelling graphic, seen on the website Visualising Data, which displays the death toll from the invasion of Iraq in a disturbing manner, using red 'dribble' lines that evoke blood running down a wall. That graph also uses an inverted y-axis."
posted by Iris Gambol at 12:07 PM on June 25, 2021 [3 favorites]

The Visualising Data example is offline now, but I found it on the internet archive.

https://web.archive.org/web/20131015034408/https://www.visualisingdata.com/blog/wp-content/uploads/2013/04/IRAQ.jpg

Visually the curves on that (at the bottom of each "drop") help reinforce that the red is the solid side of things and the white is the negative space. Compared to Chan's take, which has the black line and dots for each point on the line, which make both red and white into background.
posted by RobotHero at 12:59 PM on June 25, 2021 [3 favorites]

They can hire people who are making bad decisions out of incompetence as a cover for giving bad information to the public. Then we all sit here and say "Wow they are incompetent". It's brilliant.
posted by bleep at 1:18 PM on June 25, 2021 [1 favorite]

The problem with assuming there are mustachio-twirling villains who hire incompetent underlings as cover for their malevolent agenda is that those incompetent underlings are just as likely to do stuff inimical to their evil plans as advantageous to them.
posted by adamrice at 2:06 PM on June 25, 2021

Do they do that though?
posted by bleep at 2:42 PM on June 25, 2021

I'm not going to act like "their malevolent agenda" is unlikely just because it sounds stupid. Everything we're experiencing right now sounds stupid as hell and yet it's real.
posted by bleep at 2:44 PM on June 25, 2021 [3 favorites]

I'm slightly amazed that this hasn't resulted in an on-air retraction, let alone even a tweet saying "oops, yeah that was a bad chart to show on-air". Is this one of those 'people on twitter seem to care but nobody else really does' kind of things, so they can feel free to ignore it?
posted by TwoWordReview at 3:14 PM on June 25, 2021 [1 favorite]

Wait. HBO intern sends out integration test email, gets silently fired, gets pitied, hired by CNN as intern to make graphics … I think the timing just works, right?
posted by freecellwizard at 3:25 PM on June 25, 2021

Isn't the "surge in violent crime" really a "moderate dip in people worrying about violent crime"?
posted by kirkaracha at 3:53 PM on June 26, 2021 [3 favorites]

Yes, and since the dip was right after the start of the pandemic, it may have just been people having their threshold for "very big problem" adjusted.
posted by RobotHero at 6:58 PM on June 26, 2021 [5 favorites]