Data Visualization and Influence (Part 1)

persuasionI was really surprised to see the spike of interest my last post generated when I asked the simple question: “does visualization influence people?” few days ago. Let me start to thank you all for sending so many good and deep comments! The real value of the post in the end is in the comments, my part is definitely negligible in comparison to them (if you did not read the comments I encourage you to do so). I must confess my original post was written in the urge of getting some feedback, therefore I had not really reflected too much about the issue. Here I try to summarize some of the things I read from you guys and the new (more structured) ideas they generated.

(Note: this is the first part of a longer post. I had the feeling it was too much substance to be digested in one single post. Plus, I will try to add your reactions on this one if there will be some. So, if you have more ideas please comment below! Thanks.)

Visualization does influence people.

First of all. In retrospect I think the title of my post was a bit too ingenuous. But it worked amazingly well to attract people, so not too bad! But now in a clearer state of mind I must confess I believe my question was totally rhetorical. I think that visualization does influence people, like anything else that is involved in communication. Well, of course you never know whether a piece of information influences a person consciously or unconsciously, immediately or in a longer time span (thanks Petra for suggesting this one), deeply of superficially, and so on. I think the real problem in my original question is the too vague nature of the word influence and the lack of specification of what we mean when we ask if vis influences people (thus the variety of answers). Every piece of information influences people, you like it or not. So what?

What do we really mean by “influence”?

When we ask whether vis can influence people I think we have to better define what is “influence”. My feeling is that when we say “influence” we mean something that triggers some kind of mutation or reaction in the person who is invested by the new piece of information. But wait a moment, let’s give a look to what the dictionary has to say. The Merriam-Webster defines influence as:

Influence: the act or power of producing an effect without apparent exertion of force or direct exercise of command

Well, it seems to somewhat reinforce what I had in mind. Influence, in order to exist has to have an effect. And I think it’s on this effect that we have to focus if we want to better understand what we are talking about. Here I collected a few terms I found while reading your comments. Each one has been used in relation to how visualization influences people.

Visualization can be used as an influence tool for:

  • Teaching
  • Story Telling
  • Advocacy
  • … what else?

And can generate influence in the following ways:

  • Persuasion
  • Learning
  • Engagement
  • Opinion-Shaping
  • Empathy
  • … what else?

One thing I like about this last list is that it helps us better dissect the effect part of what we call influence (and by the way it makes it more amenable to measurement). Engagement and empathy are more related to the emotional side of it, whereas persuasion and opinion-shaping (thanks to Anthony Hamelle for suggesting this term) have more to do with permanent effects generated by “the thing” that influences.

I think an important distinction to make here is whether we aim, through visualization, at informing people, and let them do whatever they want with this information (including ignoring it) or convincing people of  something. In the end it boils down to these two main effects:

  • Increasing knowledge
  • Moving to a belief

And, while I do believe the first one is a fundamental aspect of visualization, I also do believe that it’s the second one that generates so much interest and that is totally under-researched. By the way, I think this is also what the guy at my talk meant when he asked his question.

In the end a more precise title for my post could have been: “Can visualization persuade people? I mean can we prove it?” I think this is the interesting part of the discussion. And the one that deserves more thoughts.

And the real question is not whether visualization can or cannot do that. I really think, as pointed out by some of you, that in fact visualization must have a role in the process. The open questions however are: (1) how does it compare to other means (based or not on data)? (2) how can we make this process better or worse? (3) what can we learn about visualization when framed into the context of persuasion?

How do we study visualization influence persuasion?

Well, frankly I don’t know. I mean I am not able now to specify the thing in details in the context of a blog post. But what is certain is that once we frame the question around persuasion what we will need to measure is the extent to which a belief has been induced or shifted.

I personally think A/B testing, as suggested by Michael Durwin, is neither right or wrong, it is just one option. But I do think that the goal is not only to see whether visualization can be better than other means in persuading people but also to understand the dynamics and the components that make it better or worse.

Is the data or the visualization that persuades people?

One subtle point that deserves more investigation in my opinion is the role data plays in the process. I mean is the fact that visualizations are constructed on top of data that make them more believable (and thus have stronger impact) or the visualization itself? I mean, are people more prone to accepting a belief or changing their belief when facts are supported by data? And, if so, can visualization play as a catalyst?

On a side note we should also ask ourselves if facts backed up by data tell necessarily the truth. But this is a whole other story that deserves its own blog post.

Even if this is very far from being a full design for a study, I think it’s much more precise than the things I wrote in my original pots and the things we discussed in the comments section.

Now it’s your turn! Do you want to experiment that? Does it make sense? Some of you seemed to be well inclined.

This is the first part of a longer blog post. I have more things to say on the subject but they just did not fit into one post. Plus, I want to give myself more time to think about it and, more importantly, I want to hear more from you on these new more structured ideas.

If you like this post please do not forget to retweet it and/or to comment on it. Thanks guys!

10 thoughts on “Data Visualization and Influence (Part 1)

  1. Jan Willem Tulp

    Hi Enrico,

    great post again!! Recently at SEE conference we also had a discussion somewhat related to this: as the creator of a visualization, should you create a visualization that lets the user find his own story, or is it also ok if you as a creator determines how and when a user of a visualization gets the information, like in an animated infographic with narrative elements. Some said that the latter is tricky, since it could be used as propaganda. But that of course could only be true if it does influence people.

    Now I guess you’re more talking about static or interactive information visualizations, and not necessarily about a story being told using an animated infographic. With this SEE discussion in mind, another question comes to mind: would it be possible to create an objective visualization? And what would that be? Would that be a visualization that does not intentionally influence people, but eventually may influence people, depending on the stories they find themselves?

    Looking forward to read other people’s comments on this! Thanks again for this post, a good way to start the day! :)

  2. jakob

    Enrico, I fear you are moving beyond the scope of analysing visualizations when you are looking for general psychological principles of persuasion. Then again, if you just want to find out “what works” rather than “why” at first, there might be a starting point in creating narratives, as Jan Willem suggests. The power of narratives to sway people is well documented:

    To me a narrative is but one pattern among many others you can create in a visualization, but narrative patterns are extremely salient cultural devices which make humans more susceptible to them than other, less salient patterns. This means that a world view, which as a filter totally dominates a person’s uptake of information, may be temporarily shifted through cognitive frames the person is familiar with, allowing information to pass through that would not have made it otherwise. I believe this to be the mechanism by which change of belief states can be actively facilitated.

    We should be aware though that this active use of infographics to create frames that make people susceptible to new belief states is in fact where the “fight” between Steven Few and David McCandless’ originates. To some information designers who (in my opinion misguidedly) believe that there is one and just one absolute truth hidden in data, making conscious decisions about framing the data borders on treason. To others it is the very thing they try to achieve. I think that both parties should try to understand what framing entails, before opting to choose one over the other. Something that, as your question clearly demonstrates, Enrico, is far beyond our current knowledge.

  3. @DaveAnalyst

    Persuading people is certainly a complex issue. One of my favourite examples is an experiment that found that people who had just gone up an escalator were twice as likely to give to charity as people approached after travelling downwards!

    With regards to visual persuasion, there’s a strong body of evidence that some people are ‘visual learners’ while others respond more to ‘auditory’ or ‘doing’ ( ) A big part of persuading, is explaining, and for visual learners, visuals will increase their chances of understanding you. However it’s important not to forget about the other types of learners; visuals must be seen as one of many tools needed to persuade a wide audience. Home owners even use scent, baking fresh bread to trigger positive associations in the minds of prospective buyers!

  4. @DaveAnalyst

    One additional thought – If you work in data visualisation, you’re likely to be surrounded by visual learners so it’s easy to get a skewed perspective on how to communicate and persuade others from experience alone

  5. Enrico Post author

    @Jan this is indeed slippery ground. My feeling is that it is really really hard, if not dangerous, to believe that visualizations or data can be objective. The point I want to raise is that even if the author is tries to be as neutral as possible, the data itself can offer a partial view on the phenomenon it described. I think it’s dangerous to blindly believe that whatever the data say is correct. There is a very interesting book touching the topic: Supercrunchers and if I remember well Jonathan Koomey touches the topic too in his Turning Numbers into Knowledge.

    In the end visualization is part of human expression, and the way data is interpreted too! I think it’s a better approach to let people learn to be skeptical that asking them whether the designer was neutral or not. By the way I think the same happens in journalism. And guess what? Neutral journalism or pure chronicle are just boring and don’t add up too much to the story.

    @jakob I find your discussion of narratives and frames extremely interesting. I fell like I am swimming in too far away waters, that’s not where my expertise is. But yes, I think it’s interesting when you say that cognitive frames can basically open the door to make information pass and make a difference.

    I think the debate Few vs. … ? Mmm … McCandless didn’t really participate right? Ok but the debate was not along these lines, it was more on whether it is possible to bend the rules of visual perception to make a message pass through the door (hence the frame) even if at the low communication level the design hinders (slightly) the communication.

    Anyway, as I said, I think it’s very dangerous to think the data has the truth.

    @DaveAnalyst this is certainly true Dave.

  6. Enrico Post author

    @Dave sorry you first comment was buried in the list and I overlooked it. Thanks for your links. Yeah, I am sure there is a lot of research on persuasion or influence and I agree that the visual stimulus is just one component. Another issue is whether the visualization is used as a mean by a presented or just sent off to people without a presenter. A presenter can play a very large role on the multi-dimensionality of the stimulus. Anyway, I still believe that the issue of whether data is at the root of the reason why people may believe something is a very interesting point to investigate.

  7. SellaHaAdom

    Amazing thoughts and discussion!
    How does “impact” fit in this discussion?
    I guess impact would also be easier to measure.

  8. Anthony Hamelle

    Hi Enrico,

    Great follow-up to your previous post! I’d like to concur with your views on data and objectivity. There are so many things that prevent data (and, consequently, dataviz) to be objective:
    – Cherry-picking some data over other because they don’t serve your purpose or simply because you cannot / don’t know what to make of these or how to use them;
    – Processing data (sorting them, doing additional calculations, trying to identify correlations, etc.) which implies using statistics and submitting to the inherent confidence intervals and error margins;
    – Disregarding (on purpose or out of ignorance) some data;
    – Last but not least, interpreting data with certain visual cues (colours being very important limbic triggers), choosing to compare your data to other data that make yours look good, etc.

    I fully agree with you, data fall far short of being objective. However, paraphrasing Churchill on democracy, data may be far from the truth, but they come closer thereto than anything else.


  9. Jakob

    Thanks for the kind words Enrico. Your posts, web reading and some papers I worked on recently have started to spin in my head and I’m likely to write a tentative blog post about framing, belief states and persuasion soon. I’ll keep you updated, if you are interested.

  10. Pingback: 3 Inspirational blogs for data viz : #LibTechNotes

Leave a Reply

Your email address will not be published. Required fields are marked *