Discussion: Can visualisation influence people? Can we prove it?

Enrico Bertini has triggered another important data visualisation discussion with his latest excellent piece on the Fell in Love with Data site. I wanted to extend the dialogue by presenting my thoughts from here rather than sat in a comments box.

At the crux of Enrico’s discussion is the matter of whether there is evidence out there revealing that visualisation can actually influence decision-making. He seeks suggestions for any studies that demonstrate this impact and, if they don’t exist, asks ‘why is this the case?’.

Enrico’s secondary point is to question why the concept of visualisation as a communication tool has been neglected as a topic of research in this field. I don’t entirely concur with the view that this sub-set of research activity has been neglected – any academic paper you’ve read that was concerned in some way with identifying effective methods of design will be enhancing the study of visualisation from both the analysis and communication perspectives.

I do take his point that, on the face of it, there is little available evidence about the consequences of communicated, good practice visualisation. I’m currently digging through a few years worth of papers to form one of my ‘essential resources’ post so I will specifically keep an eye out for any titles that strike me as dealing with this issue.

I believe it is important to see visualisation in the context of a wider system. Let’s consider a fundamental purpose of visualisation as being to optimise the communication of information through engaging design and accuracy of message. Once you have managed to secure sufficient attention through elegant display and have achieved the imparting of accurate information, should we also expect visualisation to hold the responsibility for facilitating a decision, triggering an action or eliciting a response of any sort?

There is a wealth of theory out there (far too much for me to have even scratched beneath the surface) about the process and mechanics of decision-making which reveals the non-linear, counter intuitive, often illogical influence that so many different variables have on the outcome such as pre-existing knowledge, preconceptions and prejudices, temporary mindset, general mood, gut instinct etc. Of course the availability of information will influence the evidential thoughts that lead to a response, even more so if it is information that has been interpreted accurately and efficiently, but as a component of the wider decision-making system it is competing with many other destabilising factors.

I would strongly urge anyone to read the book Turning Numbers into Knowledge by Jonathan Twomey. I’ve so far only got through selected chapters but it is a enlightening guide to how people do and how people should use numbers to enhance the decision-making activities involved in the act of problem solving.

Perhaps we need to consider a blend of communication tactics, deployed alongside good visualisation practice in order to maximise the communication impact. I’m thinking here about some of the elements presented in Made to Stick (by Chip and Dan Heath) and the Tipping Point (by Malcolm Gladwell) which introduce some of the different attributes believed to maximise the impact of communication such as story-telling, delivery format, oratory skills, credibility of the communicator, emotions etc.

The well-executed visualisation can only achieve so much, so lets not attach all our expectations to it alone!


JakobApril 13th, 2011 at 1:56 pm

Since my comment is stuck in moderation limbo at Enrico’s site, I hope he does not mind me repurposing it here.

You [Enrico] are right that research in the field of visual communication leaves a lot to be desired and the trend towards developing frameworks of visual language (and measuring its input) seems to still be in its infancy. (But because persuasion is so important in the VC world, there is lots of money being thrown at it.) Hence I would have to point to anecdotal evidence of this trend at first:

Edward Tufte berating powerpoint and the communication paradigm associated with it might qualify as qualitative research (as opposed to quantitative measurement of impact). He also repeatedly points out chart junk. You know you have to start with some hypotheses about visual communication before designing test paradigms anyway, so why not use his. I dare say that some of his feelings about visual communication might actually be unjustified, but his claim that (bad) visualization in NASA presentations led to bad decisions would support the notion that there is correlation between decisions and visualization.

There is surely some research done by the practitioners in the field of PR, marketing and “communication.” The success of An Inconvenient Truth to sway public opinion and thus decision makers is due in no small part to the expertise of Nancy Duarte and her team who created the original presentation. Anecdotal evidence again, but a starting point none the less. Incidentally Google Scholar nets the most promising hit for “visualization persuasion” in the paper of Sheppard, Shaw, Flanders and Burch about using visualization to combat climate change.,5

But lastly, I must point out that I disagree with your framing of “communication” as the means of persuading people. When visualizations help people understand somplex data surely that is an act of communication, no? Lots of research on impact in terms of retention rates etc has been done in education sciences. I am sure that a clearer understanding of an issue helps decision makers make better decisions, so why would you discard this perspective?

I agree with your assessment that the communication tactics geared at persuading people are a meta problem of communication. We can use methods that Meyers or the Heath brothers or Duarte (in her latest, “Resonate”) or specialists like Seth Godin employ. With the exception of Meyers most of these methods however are developed outside of academia, so there is no emphasis on a sound theoretical framework, let alone concerns of falsifiability or occams razor.

Mark from Epic GraphicApril 13th, 2011 at 9:19 pm

Nice post Andy, I enjoyed Enrico’s too.

I’m fascinated by the following point:

“Once you have managed to secure sufficient attention through elegant display and have achieved the imparting of accurate information, should we also expect visualisation to hold the responsibility for facilitating a decision, triggering an action or eliciting a response of any sort?”

Would those creating visualizations not wish to trigger a response beyond capturing attention by design? If it failed on this point, would the visualization (or communication) not be deemed empty? I guess it might depend what you mean by a response. I would classify capturing attention as a response, and would hope that the content (beyond the design) would further capture the attention. But perhaps we’re talking about the extent to which we grab their attention and move them to the point that they DO something, or at the very least change their VIEW.

I’m reminded of a quote from Tufte – “we at least want to be in the business where content matters”

But I appreciate the conflicts this can cause if we are to hold objectivity as a value or virtue in creating such visualizations. (I guess we can be thankful that these clashes of ethics or values allow us to have conversations like these).

Andy KirkApril 14th, 2011 at 9:29 am

Jakob/Mark – many thanks for your comments and furthering the discussion. You both make some great points.

Jakob – Thanks for the link to the Sheppard et al paper, I’m not sure I’ve read that one so will take a look.

Mark – You’re right to challenge the notion that I put across that once visualisation has got your attention/informed you it’s work is done. With the benefit of hindsight I didn’t articulate this very well, but you’ve clearly picked up the essence of what I’m getting at – that there are so many other factors that determine whether a decision, an action or a consequence will occur, its not just down to whether you’ve got your visualisation designed effectively (though clearly if you don’t even have this then the process is even more doomed!).