27 January 2022

Link roundup for January 2022

I have seen a lot of repurposing of fabric posters, but this one is surely the cutest:


Baby sleeping on conference poster made of fabric

Ferran Nadal-Bufi writes:

Award winning poster on peptide-based drugs to target cancer becomes play mat for tummy time 🥰

Hat tip to Needhi Bhalla.

• • • • •

In my field of neuroscience, it’s been known for some time that images of brains “lighting up” are more persuasive that tables or graphs. So a new preprint that suggests graph aesthetics influence their persuasiveness is not especially surprising.

Lin C, Thornton MA. 2022. Fooled by beautiful data: Visualization aesthetics bias trust in science, news, and social media. https://psyarxiv.com/dnr9s/

The authors say the graphs on the left were considered least beautiful, and the ones on the right were most beautiful.

Collage of eight graphs, with the ones on the left rated as more beautiful than those on the right.

A Twitter thread 🧵 from one of the authors is here.

I suspect a lot of people will flip out about this, because academics have been made very twitchy about misinformation. Any many cling to the idea that a graph should somehow be a completely neutral and impartial conveyor of data.

But graphs have always been a tool for persuasion. 

So the lesson here is not only,  “Be suspicious of beautiful graphs,” because really, you should be suspicious of any graph, but also, “It is worth the time and effort to make a nice graph, because it will spread your ideas farther. And as Seth Godin says, ‘Ideas that spread, win’.”

• • • • •

Last year, one of the big things that happened to me was getting my right eye fixed up. I’d had low vision for a few years, thanks to an undiagnosed cataract. I’m glad to be rid of it and now have the best vision I’ve had since I was a kid. But I am also greatful for the experience because it made me more alert to problems face by people with visual impairments that can't be fixed with a simple surgery.

On that note, I’m currently trying to lay my hands on a copy of:

Wu Y-H, Martiniello N, Swenor BK. Building a more accessible conference for researchers with vision impairment. JAMA Ophthalmology. In press. https://doi.org/10.1001/jamaophthalmol.2021.5613

If anyone has a copy, please email me! It seems relevant to my interests.

• • • • •

“How to present a poster at a conference?” is a short chapter (Chapter 3) in the book, How to Practice Academic Medicine and Publish from Developing Countries? A Practical Guide. (The authors chose to have a stylistic thing where not only the book title, but the chapters, and every heading ends in a question mark, whether it needs it or not.)

Open access and free to read.

Nundy S, Kakar A, Bhutta ZA. 2022. How to Practice Academic Medicine and Publish from Developing Countries? A Practical Guide. Singapore, Springer Singapore. https://doi.org/10.1007/978-981-16-5248-6_37

• • • • •

Helena Jambor reviews trends in data visualization for 2021. Topics include:

  • The ubiquity of the “viridis” colour scheme
  • Axis breaks
  • Slope charts
  • Pictograms
  • Scale bars

And more!


07 January 2022

The New York Times COVID-19 serpent

 Opinion is split on this New York Times graph:

CVOID-19 cases plotting by width of line in a spiral, with month the points on the radius.

This spiral is showing the same thing as this:

Line graph of VOIC-19 cases

A lot of people were dunking on the circular top graph. Zach Freed, who wrote, “Literally no reason to make this graph into a spiral.”

But several data viz folks perked up to defend it (here, here, here, here, here, here). (I’m not pasting in the quotes because I want to finish this post.)

And there are already remixes floating around. Amerlia Wattenberger has several.

I am more on the “detractor” side than “supporter” side. To me, this looks a lot like being different just for the sake of being different. And I don’t like that distance from center seems to be purely arbitrary.

Now, there are arguments that maybe being different is a good thing here. Neil Richards wrote, “(I)f it gets us talking about it, the chart's doing a great job!”

Yeah, but we’re talking about the graph and not the data or the ideas in the article.

It’s like when CERN announced the confirmation of the Higgs boson on slides using Comic Sans. Suddenly, people were talking about the typeface and not the data or the significance of the discovery.

The big potential advantage of a circular graph is that it is the best way to show circular data. If the big point of the graph was to explore whether COVID-19 was seasonal, yes, then this would start to make a lot more sense.

But that’s not what the article is about. Besides, even if the point was, “COVID-19 sure looks seasonal, peaking in December,” that’s n = 2 in any case.

I think Alyssa Fowers’s point that this is at the top of an op-ed article is relevant. I don’t think this is meant to show data; I think this is intended as an illustration that is based on data. I am willing to be the phrase 

“Spiralling out of control”

was uttered at some point in a story meeting. And that was the driving force leading to this graph.

Regardless of whether this plays in the “paper of record,” where people are expected to be able to spend all the time they want poring over it while sipping a hot beverage, I absolutely would not recommend this kind of graph on a conference poster.*

One of the main reasons we standardize graphs is that you don’t have to re-learn and decipher graphs anew every time. In a poster session setting, where people are busy and distracted, it’s a little rude to foist a non-standard graph on your audience if another type of graph is a much more common way of showing the data.

* Unless you were trying to show seasonality. And if you were trying to show seasonality, I would explictly annotate that on the graph.

Update, 8 January 2022: This makeover from Joe Travers is closer to what I might do. I would want some way to put the two years on the same scale, though.

Circular graph of COVID-19 cases from 2020 and 2021.

Another update, 8 January 2022: I should be writing a syllabus, not remixing graphs. Yet here I am.

CIrcular plot of new COVID-19 cases in Canada.

This is a quick and dirty proof of concept. Yes, I know the months are sloppily positioned. Yes, I know there are other things that need fixing. But I was too curious what a circular plot looked like.

Dates are a pain to work with. Spreadsheets and graphing software really, really, really want your dates to include a year and add one if you don’t. Not to mention leap years, like 2020 was.

The spiral graph was by Gus Wezerek and Sara Chodosh, though it's not clear if one or both were responsible for this particular graph.

External links

Here’s When We Expect Omicron to Peak

06 January 2022

Critique: Particulate matters

Alexandra Lai is a repeat customer of the blog. She was kind enough to share her work some time ago. She wrote, “I appreciated your critique so I am coming back for more!”

Well, when someone asks, I have to give it. Click to enlarge!

Poster titled, "From personal exposures to cell exposures Cytotoxicity and chemical composition of women’s personal PM2.5 exposures in rural China"

When I open a poster for review, there’s always that first reaction. Sometimes, it’s an overall impression that makes me go “Ooooh...” it I like it or that sound of sucking air through my teeth if I don’t. But sometimes, it’s one thing that you spot right away and you cant let go of.

It’s down in the lower right.

Headings do not just convey information about sections. They help divide a work into chunks, carving the poster at its joints. Here’s the poster with the headings at the top of each section.

While I personally have written about how I dislike changing column widths, I have very little confusion about what the sections are or what order to read the poster.

But in the lower left corner, the short heading accidentally isolates the last column,

We assume that everything under a heading  belongs to that heading. So anything that is not under a heading doesn’t feel like it belongs to anything.

Think of a heading like an umbrella.

Shadow of person holding an umbrella with the edge of the umbrella over the head, leaving on side of the body not under the umbrella.

Doesn’t just looking at that umbrella just make you want to reach in and pull it over the body? Like this?

Shadow of person holding an umbrella with the handle of the umbrella near the body, covering the body under the umbrella.

There are a few minor typographic issues that might be addressed. I strongly suspect that typesetters would not want a line to begin with a dash

– which happens in the first heading.

Similarly, I don’t think it’s common for ellipses to end one line or section…

… And use ellipses again to start the next line or section, which happens in the rightmost headings and in the text. Text lines should start with words, not symbols. 

These sorts of decisions are why professional book typesetting, even now, is at least checked by professional typesetter and not isn’t automated like Word documents. Placing line breaks and hyphens are not easily encapsulated in algorithms.

The rest of the poster has a consistent use of contrast colours, blues and oranges, that help add visual interest.

I thank Alexandra for sharing her work again, and wouldn’t mind seeing a third sometime!

Related posts