28 April 2016

Link roundup for April 2016

Lisa Rost has a nice overview of colour tools to help with data visualizations. Some have appeared on the blog before, but this is a great summary.

MarkMaker bills itself as an automated logo designer (backstory here). It’s fun to look at, but I was unimpressed with the first suggestions:

I stayed with it, trying a few favourites and deleting ones I didn’t like. I was still baffled by this suggestion after a few rounds:

I suppose it might have a certain utility in getting you out of ruts, but I’m not convinced it has much more utility than randomly picking fonts in your graphics editor. Hat tip to Doctor Becca.

DrugMonkey reports from the floor of the joint American Society of Pharmacology and Experimental Therapeutics Experimental Biology meeting:

Saw a poster with Supplemental Materials today at #aspet #expbio #eb2016 – this is where we are people.

I... wait... what? As Clay Clark asked:

On back side of the poster?

Let me make this clear:

That’s dumb. Do not do that.

We hear about viral images on the internet, but most don’t look this amazing. Even if they didn’t move, they would still be stunning scientific visualizations.

21 April 2016

Critique: Red ware

This week’s contribution comes from Scott Van Keuren. This poster recently graced the halls of the Society for American Archaeology meeting.Click to enlarge!

This a poster that has the right ideas, but doesn’t go far enough.

The title is excellent. It’s big, clean, and clear. I appreciate that this carries through the author listings, which are simpler than many posters. The logos are sensibly placed, unobtrusively, in the fine print section of the poster.

There are excellent photographs of physical objects, particularly in that critical left hand side. I would have like to have seen fewer, bigger pictures, even if that meant reducing the size of maps somewhat.

The headings also show consideration for the reader. Instead of the standard “IMRAD” headings,we mostly get questions that make it extremely clear which section of the poster is. If anything, I would like to see them bigger and more prominent. And that’s particularly valuable, because the rest of the poster sometimes leads you on a merry chase.

In the picture below, the red lines traces the order sections are meant to be read in, as I understand them:

The first column is very simple, but things take turns for the worse in the next two sections. Wrapping the scatter plots around the “What are results?” section particularly disturbs the reading flow; you have to jump a graph to get to the text, then back up to look at the scatter plots. In fact, the more I look at the poster, the less sure I am that the intended order is what I put above.

While the headings are so useful in guiding the reader, the amount I would have to read to get an answer to each question is a little intimidating. Even though I realize intellectually that the writing is not that much if it was an article, my eyes would glaze over in a poster hall.

Cutting is hard. You need to be ruthless, and you need to practice. But being concise is almost always the right way to go.

Been there, done that... but couldn’t get the T-shirt

Cathy Newman pointed out that this year’s Evolution meeting in Austin has conference T-shirts... but none in women’s styles.


Weirdly, this is a choice you have a pre-registration, months out from the conference, so it’s not as though you would make unneeded T-shirts.

In a little Twitter poll I ran, most people reported that if a conference had T-shirts, there usually weren’t women’s tees available.

Double sigh.

Come on, conference organizers. This sort of thing matters. I can’t do better than this post from Kathy Sierra:

The point is showing us that you care about more than just saving a few bucks on a t-shirt print run. That you care about ALL your users, not just the Big Burly Men.

This is partly tongue-in-cheek, but still... the t-shirts are a metaphor for – or at least a reflection of – the way the company feels about users as individual people. The shirts matter, and they speak volumes about your company.

External links

Tech t-shirts aren't sexy enough

14 April 2016

Critique and makeover: Gene sequence toolkit

This week’s poster is from Kasey Pham, and is used with permission. Click to enlarge!

Kasey writes:

I’m a student having a little trouble with my first poster presentation. I’d like to cut down the text more so that there's more white space, but I'm already having trouble keeping the story coherent.

It’s certainly nowhere near the worst I’ve seen in terms of amount of text. It seems that the main areas to edit are the introduction and the conclusion. My crack at condensing the intro was to use Randy Olson’s “And But Therefore” template:

“Every individual of a species should share a common ancestor, and this can be tested using public data, but those data are sparse, therefore we created a tool.”

I think I’m closing in on shrinking your four paragraphs down to once sentence. But I don’t know what “sparse” data means in this context, therefore I’m not sure what problem the toolkit solves.

Cutting the conclusions are more important than the intro, because that could give space around the references and acknowledgements, which currently look crowded. I wanted hack down the conclusions from five bullet points to... um... fewer. One paragraph is a challenge, but a worthwhile one.

Editing is always a bear, and the only real way to do it is with practice and constantly reminding yourself to be ruthless.

In other areas...

I’m a fan of consistent reading order, so I don’t like how the middle section switches from the reading down that you see in the left columns (the introduction flows down to methods), to reading across in the middle (Figure 1 flows across to Figure 2, then carriage returns to Figure 3, etc.). That said, the use of a horizontal line between Figures (1 + 2) and (3 + 4) is enough of a cue to prevent the reader from getting too lost.

In the Methods, it looks odd to have only the top box (“Raw XML data”) narrower than all the rest. It would also be nice for the left edge of the flowchart to align with the left edge of the text above.

Here’s a quick and dirty revision that addresses a few of these comments:

07 April 2016

The poster to publication puzzle (With stats and graphs and everything!)

How many conference posters turn into published papers?

It’s not a trivial question. A huge amount of scientific information is presented at conferences. Scientific conferences should be the places to find the “coming thing.”

But in most research fields, the importance of conference presentations pale in significance to final papers, published in peer-reviewed journals. (My understanding is that conference proceedings take on more weight engineering and computing.)

How much material from conferences is lost is relevant to discussions about the speed and efficiency of scientific communication, replication crises, file drawer problems, p hacking, and the permanence of the scientific record. It raises issues of how much you can trust what you see at conferences, and how soon you might be able to cite work that you have seen at a conference.

There is an emerging literature on this. There are nine journal articles on the topic in the last two years alone.

Because this is the poster blog, I’m most interested in how many posters eventually turn into papers. So far, I’ve found eight papers that estimated how many posters and oral talks were eventually published. Because I’m a nerd scientist, I compared the probability that posters and papers will be published. I even did stats and graphs, damn it.

Posters have a significantly lower probability of being published than talks (t7 = -7.0, p = 0.00021). Posters have about a one in five chance of being published (21.2%), while talks have about a one in three chance of being published (34.4%).

Figure 1. Square = mean, horizontal line = median; box = 50% of data; whiskers = minumum and maximum; dots = individual papers (Bakkum & Trachimowicz 2015; Daruwalla et al., 2015; Durinka et al. 2015; Janssen et al., in press; Kinsella et al. 2015; Richling et al. 2014; Singh et al. 2015; Walsh et al. 2013).

The graph doesn’t quite give a real sense of the strength of the difference between posters and talks. Every one of the eight studies found that oral presentations were more likely to be published than posters at the same meeting. Here’s one that might show that a bit better:

Figure 2. Each line represents a single paper (same papers as Figure 1).

There are several hypotheses for why this difference exists, and they are not mutually exclusive.

  • Posters may be more likely to be given by students, who might not stay in research careers. If they leave, nobody picks up writing up the project.
  • Posters are more likely to be given early in a project, which means they are more likely to “blow up on the launchpad” or never be completed. Daruwalla and colleagues (2015) found no difference in the publication delay of posters and talks, however, which speaks against this hypothesis.
  • The authors of the posters may self-select their work, presenting what they consider to be weaker or less interesting projects as posters and saving their “A game” for talks. Evidence from Sawatsky and colleagues (2015) supports this: they found that posters had a lower average rating of scientific quality than talks. (This might also explain why most people prefer talks over posters: experience has shown them talks are better.)

Another five papers calculated how likely an abstract was to become a publication, but did not separate posters and oral presentations. The average there was about the same as calculated for talks (37.7%), with ranging anywhere from 18.8% to 73.5%. The latter, for a veterinary conference, is an outlier; no other conference cracked 50%.

A few papers calculated or presented data on the time between the conference presentation and the final appearance of the publication. The delay is usually around the two year mark.

This conference presentation conversion rate might be a measure that researchers can use to convince tenure and promotion committees that they are likely to be productive in the future. If more than a third of your posters have been turned into papers in a certain time frame, you are definitely ahead of the game.

There doesn’t seem to be any hints of a relationship between the field of research or the size of the conference in whether a presentation becomes a paper. That said, a couple of papers did use the opportunity to claim that higher conversion rates were indicative of the “high quality” of particular conferences (Durawalla et al. 2015; Kinsella et al. 2015).

Of course, once I had a benchmark, I had to know if I was beating it.

I opened my CV and counted 36 conference posters in my academic career (excluding local meetings on my own campus). So far, 32 of those have been turned into publications, a very satisfying 88% conversion rate.

But... this is a little misleading. Most of the studies had a time limit on getting those abstracts published. It was usually something like four or five years. If I knocked out papers that were five years or more between poster and publication (seven of them), my conversion rate drops to 69.4%. Which is... still not bad, actually. (But that one eight year delay between poster and paper... ugh.)

The ratio of posters to papers is nowhere near one to one, though. Sometimes projects made it onto multiple posters before being published as a single paper. One project got presented as a poster seven times before I was able to seal the deal and publish the project in a journal. My 32 poster presentations yielded 17 publications.

The reverse is true, too: there are some papers that I never presented at conferences, either as a talk or poster.

This topic seems to be an unexpectedly rich vein of meta-science. I would love to see an analysis from one of the mega-conferences, like the Neuroscience or American Geophysical Union meetings.

Extra special thanks!

This week’s post would have been impossible without the generosity of the awesome Biochem Belle! She first mentioned the poster to paper puzzle on Twitter. Then she saved me a bucket of work by creating a Google document with links to the papers below. Belle:

Additional, 9 April 2016: I ran a poll on Twitter asking people how many of their posters eventually turned into talks. I got 6o votes, which is not a bad sample size. Here are the results:

I am a little surprised by these results. These results are not exactly in line with the published results I summarize in the main part of the post. Maybe it reflects that scientists on Twitter are a more awesome subset of conference goers.

Related posts

Variations on a theme: crayfish nociception


Bakkum BW, Trachimowicz R. 2015. Publication rates of abstracts presented at the 2006 meeting of the American Academy of Optometry. Optometry and Vision Science 92(11): 1069-1075. http://dx.doi.org/10.1097/OPX.0000000000000712

Chand V, Rosenfeldt FL, Pepe S. 2008. The publication rate and impact of abstracts presented at the Cardiac Society of Australia and New Zealand (1999–2005). Heart, Lung and Circulation 17(5): 375-379. http://dx.doi.org/10.1016/j.hlc.2008.02.005

Daruwalla ZJ, Huq SS, Wong KL, Nee PY, Murphy DP. 2015. “Publish or perish”—presentations at annual national orthopaedic meetings and their correlation with subsequent publication. Journal of Orthopaedic Surgery and Research 10(1): 1-8. http://dx.doi.org/10.1186/s13018-015-0203-y

Durinka JB, Chang P-N, Ortiz J. 2014. Fate of abstracts presented at the 2009 American Transplant Congress. Journal of Surgical Education 71(5): 674-679. http://dx.doi.org/10.1016/j.jsurg.2014.02.002

Dyson DH, Sparling SC. 2016. Delay in final publication following abstract presentation: American College of Veterinary Anesthesiologists Annual Meeting. Journal of Veterinary Medical Education 43(1): in press. http://dx.doi.org/10.3138/jvme.33.1.145

Fosbøl EL, Fosbøl PL, Harrington RA, Eapen ZJ, Peterson ED. 2012. Conversion of cardiovascular conference abstracts to publications. Circulation 126(24): 2819-2825. http://circ.ahajournals.org/content/126/24/2819.abstract

Janssen T, Bartels R, Lind B, Villas Tome C, Vleggeert-Lankamp CLA. Publication rate of paper and podium presentations from the European Section of the Cervical Spine Research Society Annual Meeting. European Spine Journal: in press. http://dx.doi.org/10.1007/s00586-016-4404-9

Kinsella SD, Menge TJ, Anderson AF, Spindler KP. 2015. Publication rates of podium versus poster presentations at the American Orthopaedic Society for Sports Medicine Meetings: 2006-2010. The American Journal of Sports Medicine 43(5): 1255-1259. http://dx.doi.org/10.1177/0363546515573939

Richling SM, Rapp JT, Funk JA, D’Agostini J, Garrido N, Moreno V. 2014. Low publication rate of 2005 conference presentations: Implications for practitioners serving individuals with autism and intellectual disabilities. Research in Developmental Disabilities 35(11): 2744-2750. http://dx.doi.org/10.1016/j.ridd.2014.07.023

Rosmarakis ES, Soteriades ES, Vergidis PI, Kasiakou SK, Falagas ME. 2005. From conference abstract to full paper: differences between data presented in conferences and journals. The FASEB Journal 19(7): 673-680. http://dx.doi.org/10.1096/fj.04-3140lfe

Sawatsky AP, Beckman TJ, Edakkanambeth Varayil J, Mandrekar JN, Reed DA, Wang AT. 2015. Association between study quality and publication rates of medical education abstracts presented at the Society of General Internal Medicine Annual Meeting. Journal of General Internal Medicine 30(8): 1172-1177. http://dx.doi.org/10.1007/s11606-015-3269-7

Singh A, Solanki P, Mishra D. 2014. Publication rate of scientific papers presented at the XXVI Annual Convention of National Neonatology Forum (NEOCON 2006). The Indian Journal of Pediatrics 82(1): 25-28. http://dx.doi.org/10.1007/s12098-014-1475-7

Walsh CM, Fung M, Ginsburg S. 2013. Publication of results of abstracts presented at medical education conferences. JAMA 310(21): 2307-2309. http://dx.doi.org/10.1001/jama.2013.281671

31 March 2016

Link roundup for March 2016

I have to lead with Jeremy Fox on the Dynamic Ecology blog, which tells you a big mistake almost every poster makes:

The post actually said too much text, but you get the point. (And thanks for the plug for the blog, Jeremy! Hat tip to Meghan Duffy and Pat Schloss.)

Steven Heard delves into a topic we’ve discussed on the blog before: should you give a poster or a talk?

I think the poster option is underappreciated. Because talks are seen as the default, and because they’re easier to prepare, it’s easy to slip into preferring talks without thinking carefully about the advantages and disadvantages of each format. There are major advantages to posters – especially the very high quality of one-on-one interactions they can bring – and casually defaulting to “talk” blocks off opportunities.

Steven has conferences on the brain this month, as he also wrote about how he tackles conferences as a introvert:

I like all kinds of people – one or two at a time. No matter how much I enjoy seeing my colleagues and friends, I find large quantities of them exhausting.

Ellen Lupton has a free class on poster design up at Skillshare. You need to register, but that’s all. I did so and enjoyed it a lot. If you take the class, you’re asked to design a movie poster. The student gallery is quite fun. Hat tip to, um, Ellen Lupton.

The trick of posters is to take often complex things and present them in a simple way. Brains have a reputation for being complex, so how can they be represented simply? A nice article in Nature Methods applies design to neural circuits. The figure below shows a principle (show different connections use different arrows or different colours, but not both), and a before and after critique:

Hat tip to Adam Calhoun.

David Robinson offers alternatives to pie charts.

It’s a lengthy post that is probably helpful if you are fluent in the R statistics package. (I am not, so can’t judge.) Hat tip to Michael Hoffman.

As a biologist, I’ve seen this picture of DNA many times:

What I hadn’t realized until Kindra Crick tweeted it was that this iconic scientific image was drawn by the late Odile Crick, who mostly painted nudes. It was uncredited in the original Watson and Crick paper. Like Jane Richardson (who I mentioned last month), her contribution deserves to be better known. Again, it’s a reminder that good visualizations take some skill that not everyone has, and the impact a good visual can have is enormous.

Here’s how Twitter creates its visual style. Can you articulate a style for your poster as clearly?

Today’s lesson in the importance of typography.

Hat tip to Mark Fidelman and Nancy Duarte.

And now for something completely different: a television series recommendation. While you’ve been watching Netflix original series like Daredevil or House of Cards, this one might have escaped your attention because it’s a foreign language series (Japanese).

Atelier is just a lovely series about beauty, design, craft, professionalism, and mentoring: themes that often appear on this blog. It’s subtle, often funny, and so well observed.

Oh yes, and there’s a lot of lingerie. So it’s a little more visually interesting than academia. Recommended.

(I know, lingerie shows up on this blog more often than one might expect. But it’s not always my fault!)

“Too much stuff” image from here.

24 March 2016

Lessons from Skin Wars: Have a focal point

I’ve been catching up with a show called Skin Wars on Hulu (new season coming on Game Show Network in April). It’s a competition reality show along the lines of Project Runway, Top Chef, and FaceOff: make something really cool, really fast. The cool thing they’re making in this case is body painting.

While watching the show, the judges often criticize a painting for not having a focal point. The artists make very intricate paintings, but when you step back, it’s all a confused mess. Nothing stands out.

I often see this with posters. Because posters tend to include way too much text, everything tends to turn into a uniform gray. Graphs tend to look alike.There’s few things that demand attention.

For instance, here’s a painting with no focal point:

Here’s another example of a painting, by Kadinsky, but this time with a clear focal point:

I’m willing to bet the thing that pops out is the dark circle in the upper left.

The reason is that the dark circle stands in contrast to most of the rest of the painting, which is light and has lots of straight lines and angles. Here’s another example of contrast being used to create a focal point:

It’s a contrasting colour, but a contrasting shape would work too. Imagine an unripened banana in the place of the red apple in the picture above. You’d still look more at the banana, because it is different.

Another simple way to create a focal point is with that most underused tool, white space:

There are lots of blue circles on the page, but the one surrounded by lots of white space is emphasized.

A third way to create a focal point is with lines:

Thanks to perspective, the strong lines of the train tracks, the top and bottom of the train cars, and the treetops all converge onto the vanishing point, which becomes the image’s focal point.

To use a focal point in a poster, you first need to decide what the most important thing on your poster than you want to emphasize. Once you have done that, use the three tips above (and many others besides!) to create a clear focal point on your poster.

External links

Dominance: Creating Focal Points In Your Design
Gestalt Principles: How Are Your Designs Perceived?
Designing with strong simple focal points
How to use focal points to enhance your photography
Top 25 mistakes artists make #2: not adding a focal point

Landscape from here; Kadinsky painting from here; apple picture from here; abstract from here.; mountains from here.