Hello Mr Osis, I think I am the first to try and answer this question and my answer is not comprehensive at all, it is just one example that I have to illustrate this point. There is a website called physiotherapychoices.org.au which detail systematic reviews of evidence related to therapies. They provide concise summaries for the lay public to read as well as detailed abstracts. I created a postcard / flyer based upon one of these and presented this at the physio meet and it was trying to convey the message with visuals. However due to space / crowding on my website I've had to take the picture off my website but the abstract details are there, so if you give me an email I'll send you the picture. I also did another that was related to a systematic review on the Cochrane library website. Same kind of thing and this is not much of an answer but just an example if you want to see these I can email. I need to try and get someone to pay me for these visuals, but no one to date is interested. I don't want to keep getting postcards printed, if I can sell the image that would be grand. I may see if I can upload these again later today to my website so that you can see, but space maybe an issue.

Thanks,

Deborah Hilton Statistics Online

http://sites.google.com/site/deborahhilton/

I am carrying out experiments using virtual environments for scientific visualizations tasks. I shall be able to contribute more to this topic when I am finished with it. I contacted a number of people from different disciplines in order to identify some fields that could benefit from visual analysis and gather better understanding of the problem. They all welcomed outside expertise. All of them also had different software solutions available, but they claim those solutions didn't fully addressed their "wishes". A preliminary conclusion is that those who can benefit from the visualization have difficulty expressing exactly what kind of visualization they need. This is especially true for large-datasets. In some cases, it seems to be task-related. However I don't dare as yet to claim that as a fact. By the end of January I expect to have more insight into visual reasoning for a particular domain.

Greetings Sean,

Your question, unfortunately, entails several dissertations worth of discussion, but I'll see what I can do to hit the highlights.

1) visualization is critically important to, and practically a necessity for communicating scientific knowledge. With very few exceptions, people do not think in numbers or symbols or tables. People think in pictures. There is a fundamental difference between how people's brains process what are called conventional representations (for example, numbers and letters), and sensory representations (for example graphs charts and pictures). Understanding information that is conveyed using a conventional representation requires consciously thinking about the meanings of the symbols used in the conventional representation, while information that is conveyed using a sensory representation can be almost instantly understood. This is why it is quite difficult to tell if there is a trend in noisy data from a table listing the X and Y coordinates of samples, while it is quite easy to tell if there is a trend from a scatter plot of that data.

With the additional information transfer enabled by using sensory representations, comes an additional complication. That is, sensory representations are not particularly specific. A scatter plot of X and Y values may make it easy to see a trend, but the viewer has no natural way to determine whether the trend was the only thing you intended to convey, or whether they color or shape or size of the symbols you used to plot the points, are actually additional meaningful pieces of information. This means that visualizations, even as simple as graphs, must be quite carefully constructed to maximize transfer of the intended information and minimize transfer of unintentional information.

This caveat aside, practically any scientific communication can be enhanced by the use of appropriate visual representations.

2) Despite the fact that visual representations are critical for optimal communication, there is a serious reluctance in the scientific community to acknowledge the power and benefits of visual communication. There is a small community of visualization researchers investigating optimal representations for numerous varieties of quantitative and nonquantitative data (Google VISweek for more information). However, the output from this community is rarely given the credit is due, for so fundamentally empowering the communication of other science domains.

3) For my visualization work I largely write software customized to the visualization task at hand. When possible, I prefer to leverage the rendering capabilities of packages such as POVray, or Maya, however the large majority of the “work” of visualization revolves around appropriately creating the scene to be visualized, and these canned packages simply provide a way to paint those scene descriptions into an image.

For rapid prototype visualizations, I'm quite pleased with the “Processing” language, especially in its Ruby library variant. For higher performance custom visualizations, we work in C++, Ruby, or Python, and draw using OpenGL.

4) I probably spend almost as much time creating visualizations as I do creating the data to be visualized.

5) Since one of my hats is visualization research, I don't feel a strong need for outside assistance when creating my visualizations, however if I were not my own visualization collaborator, I would definitely want one. I see far too much data being visualized poorly, and far too much potential knowledge being lost because of poor visualization, carried out by people who lack good visualization collaborators, or who believe that the charting capabilities of Microsoft Excel, are adequate to usefully convey all data.

I strongly recommend that anyone interested in effectively communicating information, both attempt to create their own visualization, and enlist the collaboration of someone who does it for a living. You are likely to arrive at a better result from the amalgamation of these 2 ideas, then from either approach alone.

I hope some of this helps,

Will Ray

A conversation like this has got to invoke the name Edward Tufte. See http://www.edwardtufte.com/tufte/ to get an idea of what's in his books.

Tufte is excellent on information aesthetics.

More seminally Jacques Bertin (http://www.amazon.com/Semiology-Graphics-Diagrams-Networks-Maps/dp/1589482611) did some of the first and still best quantitative work examining how different types of representations carry different types of information into our brains.

Recently Daniel Simons (http://www.amazon.com/Invisible-Gorilla-How-Intuitions-Deceive/dp/0307459667) has done some really nice work in the arena of how our expectations and attentive focus interact with our visual system and condition the information that does, or does not get transferred from a visual representation.

I took a class from Dr. Carr many years ago that had a lot of useful info on data visualization that I have personally used in my research. A link to a summary is at http://www.quick2degrees.com/ddata/419.pdf

Visual representations are certainly worth many words. But it is not all about professionally looking charts. Sometimes it might be worth to present a catchy cartoon to communicate a message or to stand up and draw with pencil and paper. In the end, successful communication of an idea/message can better succeed if you present it in a re-freshing way - and not using power-point and co is an interesting approach. Particularly when presenting a complex situation, you might spend days (if not weeks) to arrive at the right diagram. People from other fields might be of great support, but this certainly depends on the intended audience and if you have to step our of your comfort-zone.

For your information: you might also be interested in this very related discussion about the importance of storytelling and visualization in science (https://www.researchgate.net/post/How_important_is_storytelling_and_visualization_for_your_science).

I have to agree with Sven on this. I try to teach all our students and researchers to "tell a story" when presenting. I learned it from my mentor. A bad presentation can lose people's interest even if the data is very exciting. I've seen it all too many times. Strategic images (not pictures for pictures sake) and a good story will help the audience become engaged with your research. While our data is already very visual in nature, we use the Adobe CS Suite to make animations, images, and diagrams for our presentations. We also use Autodesk's 3D Studio Max/Maya for any 3D animations. It's cliche but often a picture is worth a thousand words. We hear that all the time with our work. I can't say enough about making sure that there's a good story to back up the data and/or visuals.

Thank you all for the answers so far. Great information! Apologies for the flurry of posts, but I would like to follow-up on many of the items discussed.

Jose, I look forward to hearing more about your work. Do you think it is fair to say that those researchers who a) benefit from visuals and b) are not certain of the direction their visuals should take, would be the ones who would benefit from outside expertise in this area?

William Ray, thank you for the in-depth discussion. I certainly agree that there are far too many examples of data being poorly visualized, or (perhaps more commonly) not visualized at all, and simply chucked into a table.

I am interested in your statement that the scientific community is reluctant to leverage the power of visuals. Do you have any ideas as to why this might be the case? It is interesting that William Playfair, who was perhaps the first proponent of charts and graphs also commented on the resistance he faced when discussing data in visual terms... and this was 200 years ago!

Mitchell and William Ray, thank you for the references. I do like the work of Tufte, but I am wondering, particularly in light of William's comment on reluctance, if these types of tools might be leaping too far ahead for many researchers. I wonder if a sort of 'baby steps' approach might receive more support from the scientific community.

Andreas, thanks for the summary of tools that you have used. I too have found that MatLab, Illustrator, and Excel are a very powerful combination. I use this combination in pretty much all of my visualization workflows.

Sven, I think you are absolutely right about presenting information in novel ways. I am more interested, however, in the use of visuals to frame data representations, rather than adding visuals for presentation's sake. Although I do think that adding something like a cartoon representation to a graph, which highlights the concept being communication by the graph, adds useful context and visual texture to an otherwise dry document.

With regards to one of your comments: Do you think that the intended audience affects the decision to seek visual expertise? And how so?

Summer, I completely agree with regards to storytelling. A professor I used to work with once created a poster presentation with nothing but visual panels (something like a comic book) which told a story about the project. When it was presented, some people sneered, some people cheered, but it was different than any other poster at the conference. I've not seen something like that since.

Sean, just to clarify, I was referring to seek any other expertise - not necessarily from visualization or communication professionals. I was trying to make the following point: if you want to address any other audience than your peers, then it is in general helpful to include representatives from the target group in your preparations.

Closer to what you apparently are interested in, it can be of great help to get professional support on any matters of communication, including visualization. However, in order to benefit from their expertise, one has to be very clear about the intended message. I have seen several cases where both parties basically talked about different things and did not end up in much (anything?) useful.

Thanks for elaborating, Sven.

Sean -- I'll try not to write a book this time -- I believe that a large part of the reluctance to embrace visual communications on the part of the hard sciences, stems from the fact that visual communication is imprecise. This is rather a "throwing the baby out with the bath water" reaction, but most hard scientists are not particularly experienced at producing visual communication, and because of their lack of experience, they regard it as "easy" and "non scientific". Coupled with the obvious issue that "easy" visual communications are subject to interpretation, while "hard" mathematical formulae (etc) are precise and absolute, and the scientist arrives at the knee-jerk conclusion that the visual communication modality is of no value.

I'm not sure that baby steps can ever get us over this mountain. Right now, you can't even convince the average hard scientist, who makes scatterplots every day, that the scatterplots that they're making _are_ visualization. The vast majority will simply ignore all of the places where they're critically dependent on visualizing their data, and tell you that visualization may be useful for communicating to the masses, but that it has no place in rigorous scientific research or dialog. It's impossible to convince a community to make baby steps, enabling you to argue by induction that further steps are likely to be valuable, when they insist on being blind to the existence of the baby steps.

The route over, I believe, involves eventual attrition out of the scientific pool, of those who did not grow up in the nintendo age. As the workforce changes from one where computers are complicated external devices on which their will must be imposed, into one where computers, and computer-generated-imagery are natural collaborators in every day life (and can't be ignored as being anything but this), my expectation is that science will focus less on the limitations inherent to visual communications, and become more open to taking advantage of the improvements that visualization brings to understanding. There will be exceptions to this route, to be sure, but for the most part, I believe that regardless of what we do to try to influence the situation, this sea-change of personal history and environment will be what eventually makes the sciences become comfortable with visualization.

Hello.

I think that visualisation is very important, especially in visualising complex, multidimensional data.

I prefer free - open source and customizable tools, so I sometimes use Gnumeric (charts), Gnuplot, R, Rapid Miner. Sometimes - MayaVi, MatPlotLib, VTK, Prefuse. All depends on what I want to show and what the "end auditor" is ;).

Regards

William Ray, thanks for your insightful explanation. Your comments have started a thought-process for me. I am now wondering if it would be a more successful approach to target and facilitate students, post-docs and young researchers in their publications and presentations. A sort-of 'grass-roots' visualization movement. I suppose it might be even better to be able to introduce this kind of material into undergraduate curriculum. Any thoughts on this strategy?

I definitely think that the student/post-doc/new-faculty end of the spectrum is easier to reach with this subject than the more seasoned researchers, but I really don't know if there's more payoff from that effort.

In many respects, it seems like they'll get there themselves, even if we do nothing, whereas most of the "old guard" won't stray from their already-well-practiced analysis and presentation methods, unless someone else dumps quite a lot of activation energy into the system, and a single success there, could have such a dramatic payoff.

Still, the undergraduate and graduate student audience is a lot more receptive, and they tend to find the idea of visualization and presentation to be enjoyable, so they're a lot more fun to teach. Right now I spend about 10% of our incoming Biophysics grad-students' course-time on visualization topics. They seem to enjoy it, they're all amused that they get to do homework for grad-school in crayon, and occasionally I get some really neat images for the office wall :-)

Zoom back from the specifics—like software—and ground yourself in basic principles and ideas around visual communication, visual literacy, drawing and cartooning. Invert your position and look at your data from the perspective of a visual artist and graphic designer. You need to have a good understanding of visual language before you can sensibly apply those principles in efficient Tufte-like ways to you data. Have a look at Chernoff faces (Herman Chernoff) and the Wong-Baker faces pain rating scale as examples of encoding information with cartoon faces as just a starter.

Have a good read of Scott McCloud's book Understanding comics: The invisible art. This book will give you plenty to reflect upon and it is a great example of comic drawing as means to encode complex information. The map as art: contemporary artists explore cartography (Katharine Harmon) is another insight into the overlap between visual arts and information representation. Once you start looking around you will discover neat overlaps between the creative arts and dry scientific data. You don't have to be Tufte to use visuals to help simplify the communication of complex ideas, effectively. Once you add the dimension of time (animation) the meaning in certain types of data come to life, makes itself visible... Like Hans Rosling's work you can see on TED.

Im a visual web designer with an animation and arts background. I can only say that visuals are critical to good communication ;-)

Apart from manually drawing or illustrating concepts when necessary, I use in combination a variety of Apple and Adobe software products.

Since 'work will always fill the time available', how long something takes simply depends on a number of practical issues. I would say this though: unless you realise early that a technique or approach does more than simply decorate then perhaps it is not worth the time required.

The point of hiring outside expertise is that you get (in theory) all the grounding that I mentioned earlier and a much faster (cheaper?) path to solving your visual needs. Meanwhile work on you own visual skills and learn from your hired help.

Dear,

in my opinion visualization is a very important topic.

When we see some presentations in conferences, symposiums, etc., it is common to see that lots of people don't care with a good presentation, or good ways to display data.

However, I have an youtube video that shows how to create a 3D animation of a scatterplot. Maybe it can help.

Please take a look at:

http://youtu.be/TDZkJ00AZug

Regards

Inkscape for 2D and Blender for 3D pictures.

Blender for animations

For instance: https://www.youtube.com/watch?v=RasemoYzsFA