Join 3,416 readers in helping fund MetaFilter (Hide)


Donut charts are mostly for decoration, right?
September 10, 2013 2:41 PM   Subscribe

WTF Visualizations is a collection of charts and graphs that make no sense. Why settle for boring old bar charts and pie charts when you can use Percentacles, Timecentages, Interferograms, the Donut Ring Explosion or whatever this is?
posted by RobotHero (95 comments total) 63 users marked this as a favorite

 
whatever this is

Two great tastes that DON'T taste great together.
posted by Greg_Ace at 2:46 PM on September 10, 2013 [2 favorites]


RobotHero: " whatever this is?"

Fuck you, clarity.

fuck you.
posted by boo_radley at 2:47 PM on September 10, 2013 [2 favorites]


This is very timely for me. I had a baffling exchange with a (talented) graphic designer earlier this year, because he would make these great-looking charts that either made no sense, or emphasized a point that was the opposite of the one readers were supposed to take away, etc. The guy was good to work with otherwise, but seemed unable to look at pictures in terms of what they meant. It was bizarre. It's interesting to see that is a more widespread problem.
posted by LobsterMitten at 2:50 PM on September 10, 2013 [1 favorite]


These charts, on the other hand, make perfect sense.
posted by weapons-grade pandemonium at 2:51 PM on September 10, 2013


I'm pretty sure I recognize the interferogram -- it's probably from InSAR imagery, in which the signal phase is compared between repeated satellite passes. There's no really great way to represent phase on a map, but the rainbow stripes are about as good as anything.
posted by irrelephant at 2:55 PM on September 10, 2013 [6 favorites]


I have no formal training in graphic design but find it incredibly fascinating. Could someone explain to me what's wrong with this one? It seems pretty good to me. How much caffeine? Yeah? How does that compare to other drinks? Got it.
posted by spikeleemajortomdickandharryconnickjrmints at 2:55 PM on September 10, 2013


spikelee...: The scale of the bars makes no sense.
posted by irrelephant at 2:56 PM on September 10, 2013 [4 favorites]


These remind me of working with a friend of mine at Bloomberg to try to figure out a good graphical way to display differences in percentage of cashflow between a bunch of different sources over time.

Additionally, my favorites are the ones that almost make sense, like the Donut Ring Explosion, which resembles nothing so much as the output of an interminable four-hour committee meeting handed off to a timid but talented young graphic designer.
posted by Navelgazer at 2:58 PM on September 10, 2013 [2 favorites]


Well, it's a bit puzzling how 0 mg/g caffeine would have a bar length at all and the apparent switch from linear to log scale between bars.
posted by bonehead at 2:58 PM on September 10, 2013 [3 favorites]


I have no formal training in graphic design but find it incredibly fascinating. Could someone explain to me what's wrong with this one? It seems pretty good to me. How much caffeine? Yeah? How does that compare to other drinks? Got it.

It's representing a range of caffeine content as a single bar. For example, the minimum amount of caffeine in black tea is only slightly more than the minimum amount in green tea, but much less than the maximum amount in green tea, so why is the bar longer? Because the average is higher? Then the coffee bar should be more than double the length of the black tea bar, but it isn't.

On preview, what irrelephant said.
posted by Noms_Tiem at 3:00 PM on September 10, 2013 [2 favorites]


I have no formal training in graphic design but find it incredibly fascinating. Could someone explain to me what's wrong with this one?

You do see the last bar on the graph where herb tea has 0mg caffeine? What do you think a bar on a bar graph that purports a zero value should look like?

I'll wait...
posted by Thorzdad at 3:04 PM on September 10, 2013 [1 favorite]


DATA SNOBS! YOU'RE ALL DATA SNOBS!
posted by spikeleemajortomdickandharryconnickjrmints at 3:07 PM on September 10, 2013 [23 favorites]


Inexpensive technologies make it possible for more and more people to communicate unclearly with each other in more and more colorful ways.
posted by PHINC at 3:08 PM on September 10, 2013 [7 favorites]


it's probably from InSAR imagery

It sure looks like a SAR polarimetry image. The colours mean different things depending on how it's been calibrated. It's used for things like crop id/monitoring, mapping of forest cut areas and burn zones, land surface structure for mapping or even mineral assessment, and soil moisture. It's a useful chart, but they're hard to interpret just from the false colour image.
posted by bonehead at 3:10 PM on September 10, 2013


See also Junk Charts (previously)
posted by Kabanos at 3:10 PM on September 10, 2013 [1 favorite]


Inexpensive technologies make it possible for more and more people to communicate unclearly with each other in more and more colorful ways.

Or, in the case of my last office job for a tech startup, have the CEO dictate all aspects of the infographics I create, including incorrect value relationships and extrapolations that don't actually exist.

But, yeah, civilians armed with copies of Illustrator is an evil that must be eradicated.
posted by Thorzdad at 3:11 PM on September 10, 2013 [1 favorite]


Donut Ring Explosion

I'll be square with you, I was hoping for goatse fingers.
posted by cortex at 3:11 PM on September 10, 2013 [3 favorites]


This one does an unintentionally great job of summarizing everything I hate about TED Talks, so it's at least useful in that regard.
posted by Copronymus at 3:14 PM on September 10, 2013 [13 favorites]


The main dashboard of a product I work with regularly is made up of donut charts, and they say exactly nothing. Very pretty and...round, though.
posted by xingcat at 3:17 PM on September 10, 2013


Are donut charts just a design in search of a function, as they appear to be, or do they actually have a specific purpose?
posted by Navelgazer at 3:20 PM on September 10, 2013


My post title comes from this one which I think is great example of some trend of trying to get the aesthetics of data visualization without having anything that warrants its use.


The interferogram does look more like an attempt to handle something that is genuinely difficult to present cleanly rather than a bungled design, so it's the odd one out.
posted by RobotHero at 3:21 PM on September 10, 2013 [1 favorite]


every time my boss says "can we include a 3D pie chart? executives love pie charts," for some non-percentage visualization i come back to this page to find inspiration for the unsolvable quandry i've been tasked with. i gnash and i wail and i make a non-pie chart that still meets the unspoken requirement of "make it pop!" while doing my level best to make sure what i'm presenting wouldn't belong here.
posted by radiosilents at 3:23 PM on September 10, 2013


"can we include a 3D pie chart?"

Send your boss this.
posted by Kabanos at 3:30 PM on September 10, 2013 [27 favorites]


This is the way the world ends
This is the way the world ends
This is the way the world ends
Not with a ! but a ;
posted by It's Raining Florence Henderson at 3:33 PM on September 10, 2013 [2 favorites]


I can hardly wait for The Donut Ring Explosions' "Best Of" anthology coming soon on Rhino Records.
posted by Cookiebastard at 3:38 PM on September 10, 2013 [2 favorites]


Oh to have the img tag so I could represent how much I love this in donut chart form (hint: the total would add up to more than 100%)
posted by TwoWordReview at 3:47 PM on September 10, 2013 [4 favorites]


This inspires me to make the most useless graph possible with the next bit of data I come across. Excel has a Bar of Pie chart that looks deliciously unreadable.
posted by maryr at 3:51 PM on September 10, 2013 [1 favorite]


PS: I recommend Googling "worst graph ever".
posted by maryr at 3:52 PM on September 10, 2013 [1 favorite]


The greatest trick the Devil ever pulled was convincing the world to use PowerPoint.
posted by It's Raining Florence Henderson at 3:57 PM on September 10, 2013 [7 favorites]


Are donut charts just a design in search of a function, as they appear to be, or do they actually have a specific purpose?

You can nestle donut charts (with a pie chart in the middle) to show a change in percent distribution over time, across regions, or whatever. Whether or not that's the best way to do such a thing is debatable, but it can be done.
posted by Kid Charlemagne at 4:00 PM on September 10, 2013


The worst graphs are the ones that take 10 minutes to create, are based off of two spuriously correlated data points that have been proven as such, where a highly respected coworker in another department regresses one on the other over a short period of time where they fit, where the conclusion drawn from the data and the representation is unbelievably false based on the data sample sets. Then, to boot, they then circulate the insight around senior leadership, who require two weeks of work to put exorcise that demon back in to the lands of erroneous conclusions.

The truly worst graphs don't confuse or tell you nothing - they make you waste your time.
posted by Nanukthedog at 4:01 PM on September 10, 2013 [4 favorites]


This one does an unintentionally great job of summarizing everything I hate about TED Talks, so it's at least useful in that regard.

The best part is that it starts out shit talking the quality of the graphics during TED talks. Surprised the were able to operate the computer, what with the plank in their eye and all.
posted by Homeboy Trouble at 4:01 PM on September 10, 2013


Donut Ring Explosion

I'll be square with you, I was hoping for goatse fingers.


Something like this, perhaps?
posted by Greg_Ace at 4:02 PM on September 10, 2013


They're all kind of pretty as long as you don't attempt to figure out what they mean.
posted by tommasz at 4:03 PM on September 10, 2013


Are donut charts just a design in search of a function, as they appear to be, or do they actually have a specific purpose?

You can also dunk doughnut charts in a cup of coffee.

Pie charts are evil, doghnut charts are only slightly better, but only if you are using the width of the path as well as the distance between paths to record related data. Otherwise it is a pie chart being misused - and pie charts should never be used.
posted by Nanukthedog at 4:04 PM on September 10, 2013


People don't understand what Venn Diagrams are for....
posted by Apropos of Something at 4:09 PM on September 10, 2013 [15 favorites]


I think that's a koan.
posted by vogon_poet at 4:11 PM on September 10, 2013 [1 favorite]


In all fairness, THIS SHIT IS HARD!

Look at this "how it should have been done" graph from Junkcharts. Sites with higher ratios of female users are all going to be squeezed into a tiny strip between 0 and 1 while sites with more male users are going to get spread all over hell's half-acre. In an extreme case, a site that has 9999 female users and 1 male user will be one unit away from the 1 to 1 mark and and a site with the opposite gender ratio will require me to scroll to the right for about 5 minutes.

A better way to do it would be to have the bottom axis be something like "Gender Ratio" and go from 100% male to 50/50 to 100% female.

This shit being hard in now way excuses exploding donut charts.
posted by Kid Charlemagne at 4:11 PM on September 10, 2013 [1 favorite]


Apropos of Something: People don't understand what Venn Diagrams are for....

The message is obvious! That diagram's saying that appropriating Native American images as team logos isn't really acceptable...but it's done anyway.
posted by Greg_Ace at 4:16 PM on September 10, 2013 [15 favorites]


Lies, damned lies, statistics and charts.
posted by islander at 4:24 PM on September 10, 2013 [1 favorite]


People don't understand what Venn Diagrams are for.

Clearly that's a Mastercard Logogram.

this is a Glenn diagram
posted by cortex at 4:24 PM on September 10, 2013 [2 favorites]


this is a Dern Viagram
posted by It's Raining Florence Henderson at 4:29 PM on September 10, 2013


Technically, of course, the "allowed" list should just say "all things not on the prohibited list."
posted by Apropos of Something at 4:31 PM on September 10, 2013


Seriously, that Potomac Area Football Team venn diagram is like is like a grain of sand in my brain which will eventually produce a pearl.
posted by Navelgazer at 4:40 PM on September 10, 2013 [3 favorites]


This one does an unintentionally great job of summarizing everything I hate about TED Talks, so it's at least useful in that regard.

Ha. It's almost like it kept following the donut chart labelled "awfulness" all the way around until it became a surreal illustration of a deeper truth.
posted by graphnerd at 4:48 PM on September 10, 2013


I can't visualize what some of these people were thinking.
posted by double block and bleed at 5:02 PM on September 10, 2013 [1 favorite]


I'm pretty sure I recognize the interferogram --

Me too! It is a piece of fabric torn from a Hypercolor heat sensitive t-shirt I sold at a garage sale in the early 1990s.
posted by MoonOrb at 5:06 PM on September 10, 2013 [2 favorites]


Visualize weird piece.
posted by It's Raining Florence Henderson at 5:07 PM on September 10, 2013


I went to an Edward Tufte seminar once, where he sneered at the "chartoonists" who made infographics for newspapers.

The visualizations in this tumblr... most of them could only aspire to one day grow up to be chartoons.
posted by Zed at 5:08 PM on September 10, 2013


These are CHUDtoons.
posted by It's Raining Florence Henderson at 5:12 PM on September 10, 2013 [1 favorite]


Sites with higher ratios of female users are all going to be squeezed into a tiny strip between 0 and 1 while sites with more male users are going to get spread all over hell's half-acre.

The traditional thing to do would be to use the odds or logit (relabeled as you suggest) as the abscissa.
posted by Ms Vegetable at 5:16 PM on September 10, 2013


I feel dumb. What's wrong with this?
posted by LogicalDash at 5:17 PM on September 10, 2013


You really need to get that abscissa drained.
posted by It's Raining Florence Henderson at 5:18 PM on September 10, 2013 [1 favorite]


LogicalDash - first, the GREEN 185 is twice the height on the y-axis as the larger Supplier and Transport numbers. Only the GREEN seems to really correspond to anything on the y-axis.

Second, it is ugly.
posted by maryr at 5:24 PM on September 10, 2013 [1 favorite]


I am so, so thankful that I was born in the era where computers were just starting to be used for shit like this and people still cared about representing information correctly. I had courses in middle and high school that included classes about what graphs are best for which kind of data, how to write for readability, and all the other things that seemed to get hand-waved away nowadays with "the computer will fix it for you" mentality.

Now if you'll excuse me, I need to go wash the blood out of my eyes after viewing that website.
posted by backseatpilot at 5:29 PM on September 10, 2013 [2 favorites]


Only the GREEN seems to really correspond to anything on the y-axis.

Third, this does illustrate a weakness with stacked bar graphs. The red, black and blue are all identical heights for each year, but it's easy to miss that because they aren't lined up and that makes it a little harder to compare them.

Fourth, and now I'm getting nit-picky, even though two years pass between the first two stacks and five years pass between the next two stacks, they are evenly spaced horizontally, which could misrepresent trends in Green costs if that's what you are looking for.
posted by RobotHero at 5:40 PM on September 10, 2013 [1 favorite]


Damn -- I was TRYING to fuck this up when I made it a few years ago -- I didn't try hard enough.
posted by Devils Rancher at 5:46 PM on September 10, 2013 [13 favorites]


Arrragh! This shit drives me wild. Some of it is just stupid I-can-make-a-pretty-picture for my presentation in Excel or whatever, but much of it is deliberate manipulation to misrepresent facts.

Suppose there's an article using that stacked bar graph referenced above. Lots and lots of numbers trying to claim how expensive it is to be energy efficient (green.) People generally don't really RTFA, because it's boring and hard to figure out exactly what's being said. So they turn to the chart. Look, the green section is really big it must be really expensive. Half the time what's being depicted is irrelevant, off scale, skewed, or generally borked. It takes energy to figure out if a graph is depicting what it intends to for most of us, and for many it isn't intuitive.

Here's a primer on How to Lie with bar graphs. Here's how the Republicans did it. And here are evil pie charts.

Data vis is a pretty good website, if you're really into bad graphs
posted by BlueHorse at 5:56 PM on September 10, 2013 [7 favorites]


Send your boss this.

No, send your boss this.
posted by Horace Rumpole at 6:01 PM on September 10, 2013


From the "intentionally bad" pile:

The principles of Eisenstein's relativity are too well known today to require explanation. The humblest reader of this book can undoubtedly give a clear and concise account of such relativistic phenomena as the Michealson-Morley experiment, the Lorentz-Fitzgerald contraction and the Dow-Jones Average. The familiar World-Line Diagram (below) sums it all up.

Tom Weller, Science Made Stupid (1986)
posted by dhens at 6:08 PM on September 10, 2013 [4 favorites]


I feel dumb. What's wrong with this?

The sizes of the bars are in no way proportional to the amounts; in the middle bar for instance, Green (£212) is the largest bar. This bar is roughly the same size as the bar for Production (£540 - actually more than twice as much as Green), over twice the size as the bar for Transport (£340 - actually 50% more than Green), and something like four times the size as the bar for Supplier (£219 - about the same as Green). Even though it seems to be emphasizing Green, it actually understates the significant increase projected for Green -- a 50% increase looks like a 20% increase.

While it looks like a bar chart, it is in no way a bar chart. Here's a comparison between the original and a (less shinily-formatted) accurate version.
posted by Homeboy Trouble at 6:21 PM on September 10, 2013 [3 favorites]


These are fucking amazing.
posted by Saxon Kane at 6:26 PM on September 10, 2013 [1 favorite]


Homeboy's made the bars proportional, but the chart still lies--between lines one and two is a two year gap, between lines two and three is a five year gap, as RobotHero mentioned above. If you were talking company growth, it almost looks like there's been change over equal amounts of time. Intellectually, you may realize the time gap is different, but the overall impression is that the growth is even from year to year.

Here's a history of dishonest Fox news charts. *gasp*

Charts and graphs are fun!
posted by BlueHorse at 6:46 PM on September 10, 2013 [1 favorite]


WTF Visualizations is a collection of charts and graphs that make no sense.

At my workplace we call these "infographics".
posted by benzenedream at 6:55 PM on September 10, 2013


This one is terrible. It is supposed to be from an info graphics firm, and yet the proportions of each pie segment are inversely proportional to the size of each segment... Who got paid to make this?

I have seen some awful, awful graphs. I spend a lot of time helping others get data in order for grant applications, and the things I have seen... But damn if this site doesn't make the crap I deal with look tame.
posted by caution live frogs at 7:03 PM on September 10, 2013


Is this where I get to link to rap lyrics represented as mathematical graphs? I think it is.

(if only there was a chart or graph so I could know for sure)
posted by Mchelly at 7:38 PM on September 10, 2013 [2 favorites]


This image referring to The Disciplines of User Experience is really something special. I think I'll either get it laminated for the love of irony, or send it to people I hate.
posted by quiet earth at 8:01 PM on September 10, 2013 [1 favorite]


This one is terrible... Who got paid to make this?
http://www.boostlabs.com/
Their web site is a mess, as you might imagine.
posted by bhnyc at 8:03 PM on September 10, 2013


Workflow all the things.
posted by flyingsquirrel at 8:47 PM on September 10, 2013


An oldie but a goodie: still the best pie chart ever.
posted by George_Spiggott at 8:59 PM on September 10, 2013


This one is terrible. It is supposed to be from an info graphics firm, and yet the proportions of each pie segment are inversely proportional to the size of each segment... Who got paid to make this?

I stared and stared at this trying to figure out the relationship between the pie wedges and the numbers until finally I realized that it's actually mislabeled. The numbers are in backwards order with respect to the wedges they point to--the outer pink should be SEO and the inner red should be Online Media Buys.

So apparently this graph was so amazingly intuitive that they managed to forget which wedge was which during the process of creating the graph. Now that's some design!
posted by equalpants at 10:04 PM on September 10, 2013


backseatpilot: "I am so, so thankful that I was born in the era where computers were just starting to be used for shit like this and people still cared about representing information correctly. I had courses in middle and high school that included classes about what graphs are best for which kind of data, how to write for readability, and all the other things that seemed to get hand-waved away nowadays with "the computer will fix it for you" mentality."

Yeah, except for the fact that all the people demanding these graphs be made are the people who grew up taking those classes and are now the adults running things.
posted by symbioid at 11:25 PM on September 10, 2013


In all fairness, THIS SHIT IS HARD!

Well the worst part of that chart is that they don't label the x-axis so that you can understand which side is which gender. Until you see Reddit sitting there all the way by itself on the right anyway. By quite a remarkable distance it must be said.

Which in any case means that they should have labelled the x-axis 'female-male ratio'
posted by Hello, I'm David McGahan at 5:11 AM on September 11, 2013


Which is to say that the skewing of the graph has more to do with keeping Reddit on it than anything else, which is actually visualising something about the gender skew of the sites they were putting on the graph. So yes, that's how it should be done.
posted by Hello, I'm David McGahan at 5:14 AM on September 11, 2013


But yes this stuff can be hard. Was in a meeting a few weeks ago where everyone was freaking out about a revenue trend graph until someone realised that the x-axis had time going from right to left rather than left to right.
posted by Hello, I'm David McGahan at 5:19 AM on September 11, 2013 [1 favorite]


Which is to say that the skewing of the graph has more to do with keeping Reddit on it than anything else, which is actually visualising something about the gender skew of the sites they were putting on the graph. So yes, that's how it should be done.

No, the skewing of the graph is due to them using a ratio of numbers of users in each category. So that 3.0 that Reddit approaches actually means that it isn't all that far off from the inverse 0.333... that Pinterest approaches. Except that the distance from 1.0 along the abcissa is 2.0/0.666... = 3 times greater for Reddit than Pinterest. Hence, it is misleading.
posted by indubitable at 5:37 AM on September 11, 2013 [3 favorites]


In my experience, doughnut charts are what you make when someone asks you for a pie chart, and you can't say no, so you compromise. I made some nice doughnut charts for everyone's posters this conference season, but then they went and turned them into weird ovals, and I cried inside. I was like 'But nothing in the chart now bears any relationship to the data!' and they were all 'But I had a rectangular space left on the poster, so I stretched it'.

I thought the bad charts problem was one confined to the mostly-innumerate business and journalism industries, and then I started working with scientists. Aaaaaaaaaagh.
posted by Acheman at 6:12 AM on September 11, 2013 [1 favorite]


Sorry don't see it. You can express it as a ratio a percentage or whatever, which are basically the same thing expressed in different ways. The graph is only a ratio which sez that for every 3 men on reddit it's getting somewhat uncomfortably close to zero women. They chose to move the centre point, which sure that's debatable, unless you think they are trying to represent the data they have I suppose.
posted by Hello, I'm David McGahan at 7:16 AM on September 11, 2013


Which in any case is more boring than looking at the twitter-yelp-myspace-pinterest-instrgram nexus. What's going on there is an actual interesting question.
posted by Hello, I'm David McGahan at 7:24 AM on September 11, 2013


The message is obvious! That diagram's saying that appropriating Native American images as team logos isn't really acceptable...but it's done anyway.

If by appropriate you mean use with blessing and permission then sure, that statement makes sense and I'll stop with the derail derailer here.

posted by RolandOfEld at 7:35 AM on September 11, 2013 [1 favorite]


Sorry don't see it. You can express it as a ratio a percentage or whatever, which are basically the same thing expressed in different ways.

But accounting for how people will interpret something is part of decent design, and one thing that's pretty well known about people is that in a null context they're going to look at stuff on a line and interpret it linearly. Ratio-on-a-line and percentage-on-a-line will plot values with different distributions; the difference matters a lot, like the difference between a generic linear axis scale and a generic log axis scale matters.

That it's possible to carefully decode the meaningful numbers from a chart is not the same thing as that chart presenting those numbers well, basically.
posted by cortex at 7:58 AM on September 11, 2013


What indubitable is saying is it hinders comparison between ratios on either side. If you have a site with 3 men to 1 woman and a site with 3 women to 1 man, then the former is positioned 2 units away from the even ratio, while the latter is positioned only 2/3 of a unit away. Even if you had a site with 99 women for every man, it could never be more than 1 unit away from the "even" line, but a site with 99 men for every woman would be 98 units away.

But if you present it on a scale from 0% men to 100% men, (Or 100% women to 0% women) then the 3 men to 1 woman site is positioned at 75% men line, and the 3 women to 1 man site is positioned at 25% men line, both are the same distance away from the centre of 50% which makes the comparison possible to intuit from the graph.
posted by RobotHero at 8:11 AM on September 11, 2013


'But I had a rectangular space left on the poster, so I stretched it'.

Oh noooooo ...



If we're sharing stories, we had a vector-art map and gave it to a graphic designer to make it fit our branding. He came back with the roads, the borders, and the rivers all the same colour and weight so they were indistinguishable. I guess he was only thinking in terms of the impression when you looked at the map as a whole, he wasn't thinking at all about how annoying it would be to read this map. We didn't print it like that, (I convinced him to go back and adjust the line weights) but it was a memorable way for a map to not work very well as a map.
posted by RobotHero at 8:42 AM on September 11, 2013 [1 favorite]


I knew I'd seen the 'whatever this is' before and was completely unsuccessful in googling it last night. It's actually a link to an epically long info graphic put out by GW's Master of Public Health program. You can see the full thing on their blog.. The full infographic is about three times longer than the snippet on WTF Visualizations but it turns out to be done by public health students, not design professionals, so there's only so much judging that's fair.
posted by librarylis at 9:26 AM on September 11, 2013


Women at age 18—martini glass; age 34—baby?
posted by fontophilic at 10:51 AM on September 11, 2013 [2 favorites]


BEST GRAPH
posted by tigrrrlily at 11:21 AM on September 11, 2013 [1 favorite]


It's actually a link to an epically long info graphic put out by GW's Master of Public Health program. You can see the full thing on their blog.

Mm hmm.

The spaghetti results from they are trying to
1) have a series of bar graphs
2) they want to order each of those bar graphs by height, but
3) they want you to be able to follow one country from one bar graph to the next even though the order is changing

For the connecting the gross national income per capita to health expenditure per capita, the spaghetti kind of works. The two are mostly correlated so any country that bucks the trend will stand out in the spaghetti.

The population bubble chart is annoying. I suppose they opted for bubbles instead of bars because unless they used a log axis, India would need to be 400 times the length of Mongolia. But since the bubble's position doesn't seem to correspond with real info, I feel they could have positioned them more in line with their order on the following bar graph and reduce the "what is this mess!?" reaction that you get when this is the first thing you see.
posted by RobotHero at 3:02 PM on September 11, 2013


Not at all sure I understand why this one is a problem, it looks to me like it was supposed to be funny in the first place.
posted by hap_hazard at 3:49 PM on September 11, 2013


I thought the bad charts problem was one confined to the mostly-innumerate business and journalism industries, and then I started working with scientists. Aaaaaaaaaagh.

I try so, so hard to convince my coworkers that you can change the colors in the Excel graphs. For example, your low temperature could be blue and your high temp could be red. Intuitive! But noooo, red-blue-green-purple it is for EVERYTHING. And why be consistent between two graphs?

Also, my boss is apparently color blind, which is infuriating when I make such BEAUTIFUL RAINBOW GRAPHS.
posted by maryr at 9:25 PM on September 11, 2013


Also also, apparently everything in a patent has to be black and white, so good luck with that dot plot of 10 different samples now that Excel has eliminated pattern filling!
posted by maryr at 9:26 PM on September 11, 2013


The two are mostly correlated so any country that bucks the trend will stand out in the spaghetti.

Look, if they'd even used straight lines, it would have been better, because then you'd have been able to see from the slope of the graph whether a country was spending disproportionately more or disproportionately less on healthcare per capita as compared to GDP. As it is, there's just a tangle of lines and you have to read off each country separately - a simple table would have accomplished the same, and still been easier to read.
posted by Acheman at 2:39 AM on September 12, 2013


I just found my own WTF Visualization in the wild today. This is just a snippet of the world map (of countries which have a First Lego League). Aside from the giant incomprehensible mess of labels and lines, I liked that the Netherlands has moved to Scotland; Scandinavia is now a country on the border of Sweden and Norway; Ireland and Iceland are (deliberately?) ambiguously labeled.
posted by Joh at 4:11 PM on September 12, 2013


Andrew Wheeler tries to give the opposite experience here, how to avoid pitting form against function. I like the idea of thiefing colour schemes from famous paintings.

Acheman, you've got a good point with the straight lines could have been better. Like if you look when India and Ghana switch positions, India has to make two extra turns to avoid them crashing head-long, while if they were just straight lines they would have just been an X. This scheme of rounding angles off to the nearest 90 degrees is something they use in subway maps (Though they would usually allow diagonals at 45 or 30 degrees) but it's misapplied here if the lines need to make more turns.
posted by RobotHero at 9:51 AM on September 13, 2013


Those links actually gave me heart palpitations. From trying to understand abstract shapes. Yeesh.
posted by psoas at 9:40 AM on September 17, 2013


They're polling for whether it's better to source where the visualization comes from or keep it anonymous

I think I am glad to have seen the whole version of the graph librarylis linked to. It made me think deeper about how this jumble had come to exist.
posted by RobotHero at 11:03 AM on September 20, 2013


« Older John Wooley's leaked Market Research Report...  |  In the summer of 2000, the top... Newer »


This thread has been archived and is closed to new comments