Have a look at this short talk by David McCandless on data visualization:
There are a number of interesting things for argumentation theorists to think about here. For one, if McCandless is right then clearly what he says is wind in the sails for those who rate argument diagramming highly among our various tools of analysis.
While watching this presentation I also found myself wondering if McCandless’ technique might provide aid and comfort to the cause of those who believe in visual arguments too. To be clear, I don’t think that any of the visuals he presents here is an argument. He makes visual statements, sure, and at times draws inferences from them, but that would make his arguments (in my book at least) arguments with visual elements–not visual arguments per se. Still I found myself wondering if maybe purely visual arguments might be a possible innovation that could come from the kind of work McCandless is doing, somewhere down the line.
RAIL 
Interesting connection! I must ask, what do you mean by “visual arguments per se“?
Well, you have folks like Groarke and Birdsell who talk about “purely visual arguments”. For them, such an argument would be made up only of visual elements, and would be capable of displaying an inference from premises to conclusion absent the use of any linguistic component. Words, they say, aren’t necessary. Pictures alone will do. My usage of “visual arguments per se” is meant to respect this idea.
I take it to be controversial whether purely visual arguments after the fashion of Groarke and Birdsell actually exist. That said, I think it fairly obvious that a picture could be part of an argument if used alongside linguistic components as usually happens.
If that, then I agree. 🙂
“Arguments consisting only of images” cannot but contain either a metaphorical – or loose – sense of ‘consisting only of’ or of ‘arguments’. I cannot see arguments but as linguistic expressions of an inference, even if, at a further step, we ‘transform’ some of its material into visual expression. Then, however, the ‘visual’ would only be a secondary means of expression which must be understood as secondary. The argument would not, strictly speaking ‘consist of’ it.
“Knowledge compression”.
While this is particularly useful to identify first order trends (say the overall slope of a curve) or simple relationships, high resolution detail is necessary to discern the true story.
Take for instance the Keeling Curve, which illustrates measurements of concentrations of atmospheric carbon dioxide since the late 1950’s at Mauna Loa, Hawaii. It is often cited as evidence of anthropogenic forcing of atmospheric CO2 levels in an upward trend.
http://earthobservatory.nasa.gov/IOTD/view.php?id=5620
The first order trend of CO2 levels is most certainly upward. However, a high resolution view shows regular, yearly oscillations between low and high levels of CO2. These seasonal fluctuations can be accounted for when CO2 levels in the atmosphere are mitigated by the CO2 uptake of plants during spring and summer in the N. Hemisphere. So, every spring/summer there is a downward second order trend in atmospheric CO2 levels. An even higher resolution view reveals daily oscillations of CO2 levels attributed to plant and soil respiration.
If you have never seen this curve, especially, if I am pretty sure you will never look it up I can safely make up any argument that I want about CO2 levels. Further, I can slice,”clean”, extrapolate, manipulate and visualize data so that I am only showing you the second order trend or whatever false “trend” I want to show you. Furthermore, I can map any number of other orbital changes or sunspot curves over top of it to provide you with “context” that supports my argument. If you decide to actually look it up, understand it and form your own opinion… well then I’m probably screwed.
What the information communicates is largely dependent upon the intent of the person compiling the data and the resolution view of the data that is used. So instead of “let the data set change your mindset” might we “let the ENTIRE data set change your mindset”. This brings me to my second concern which was touched upon in another post here on RAIL.
https://railct.com/2011/01/03/defense-against-the-dark-arts/
Citing the sources of data sets might not be enough. Fact checking is laborious. Most are prone to misdirection and laziness when it comes to verifying sources. Most don’t have the time.
I’m not sure what would suffice aside from publishing the entire data set, how and by whom it was compiled, and the method, software and/or calculus by which the visual came to be rendered. This is impractical. Clearly there needs to be more of a discussion regarding ethical practices & protocols among data visualizers and journalists.
McCandless is convinced of the power of data to change the world for good. Thankfully, he is aware of bias and seems to be a decent person whose motivations are in the right place. Data visuals are compelling but they are also dangerous because they are so direct, inspirational and seem to provide instantaneous validity to arguments.
Visuals seem to have the power to provide the viewer (and obviously the person making the argument) with the feeling that their view is valid regardless of the truth of their premises or the cogency of their beliefs. The DJIA curve changes every day and every day somebody’s analysis of the previous days close and their investment strategy for the next day turns out to be completely wrong. Forming beliefs from data, artfully visualized or not, is a dangerous business and is no more elegant than pointing to your neighbor’s car and forming an opinion of their character based upon what condition it is in.
To his credit McCandless makes mention of bias and context dependent interpretation but perhaps he ought to be more careful when he talks about “cleaning” and “engineering” data. I find myself wondering if what McCandless is doing is more art than rigorous analysis. Indeed, what he does with numbers is both elegant and beautiful.
I suppose I would have appreciated more discussion of the potential pitfalls of data visualization and less glorification of data as more than just the tool that it is. No doubt, as data visualization software becomes easier to use and more journalists become “data visualizers” these concerns will intensify.
[…] 5, 2012 by Steve Readers of RAIL might remember this chestnut from two years ago on infographics and visual argument. That post featured a TED talk by David McCandless. Though I’m tempted, I’ll refuse […]