Skip to content
14 June 2011 / erikduval

Goal oriented visualizations?

Some of the fine fleur of information visualisation in Europe gathered in Brussels today at the Visualizing Europe meeting. Definitely worth to follow the links of the speakers on the program! Twitter has a good trace of what was discussed. Revisit offers a rather different view on that discussion than your typical twitter timeline.

In the Q&A session, Paul Kahn asked the Rather Big Question: how do you choose between different design alternatives for visualisation? Both Moritz and David answered that they rely on “Ego Driven, self centred” design😉 , i.e. they consider the options and decide. No user studies. No feedback from peers. (They do get quite a bit of community reaction after they publish their visualisations and Moritz does do careful tracking of how people actually use them.)

In their case, that answer is sort of OK. But, in general, it is less than satisfactory for me: in usability engineering, “the user is always right” and “the developer is not the user”. I am not so sure that the information visualisation community sees things that way: there is a bit of the “we are artists and we are too sophisticated for ordinary folks” and “if they don’t understand it, they should take an information literacy course” attitude there. (Or should I include myself in that crowd and write “here” rather than “there”?)

Yet, if visualisations are to have any effect beyond the initial “wow” factor, it would be useful to have more clarity on what the intended goal is and how to assess whether that goal is achieved. Take Moritz’s wikipedia visualization for instance: admittedly, this looks good – it is actually beautiful. But Moritz mentioned that he thinks of himself as a “truth and beauty operator”. I wonder what kind of truth this visualization reveals? Do I use or read wikipedia differently afterwards?

What interests me these days is how we can connect visualization not only with meaning (truth, if you want), but with taking actions. This is very much in a “quantified self” approach, where for instance a visualisation of your eating habits can help you to lead a healthier life, or where a visualisation of your mobility patterns can help you to explore alternative modes of transport, etc. Such visualisations would be successful if they trigger the intended behaviour. That would be something one can measure, as in “people smoke less when they use this visualisation” or “people discover new publications based on this visualisation” (we’re actually evaluating such an application) or “people run more using this visualisation” etc.

It would be really useful if we could draw up some guidelines to design effective goal oriented visualisations. As an example, I guess it’s kind of useful to be able to visualise progress towards a goal – or lack thereof. If you want to run further, a visualisation can help you to assess whether you’re making progress. Or if you want to spend less time doing email, a simple visualisation can help. Another guideline could relate to social support, that enables you to compare your progress with that of others. Etc.

Do you know any good examples of information visualisation that support a specific goal? Do you have any suggestions for design guidelines that can help to create more and more useful such visualisations?

(And did I mention that I need to devote at least an afternoon soon to trying out Impure?)

13 Comments

Leave a Comment
  1. Marcus Specht / Jun 14 2011 11:24 pm

    Hei Erik, I like the informative art approach. They really use visualisation for embedding ambient displays in the user environment with a specific communication or even learning goal. (http://www.sics.se/fal/projects/infoart/) We follow that up in the work of Dirk Boerner who uses ambient displays for aiming at a change in power consumption behaviour of people.

  2. xdxd_vs_xdxd / Jun 15 2011 1:02 am

    Hi Erik! Well, actually this is the whole point! (at least for me and for those that are the things which i find important)
    today it was a really fast presentation and i am not sure that i was able to get the whole message through, but for FakePress it is all about process and change, mutation. In more than one direction.
    On one side, research on what has already changed, and produce for that: so this is where the whole ideas of ubiquitous, pocketable, wearable, walkable information and interactions are born, being triggered by an already_mutated human being which actually already has several new sensorialities which are externalised onto devices and technologies.
    On the other side, research on what new processes can emerge, and produce for that: so we think about the experiences that shape/fill our lives, like shopping, learning, relating, working, and trying to bring up new scenarios which build on the knowledge about what already happened (loads of accessible data, accessible networks and technologies, accessible global relationships, accessible global processes, accessible possibilities to let hyper-local processes explode into globality… etcetera) to experiment the “next-steps” of it all… and this is where projects like Squatting Supermarkets are born.
    And there are suprises in both approaches: when for example you build something for a specific purpose or profile and the people which represent or embody that purpose/profile actually find the “thing” you produce really useful and engaging, but for a really different purpose than te one you had envisioned.
    This, for example, happened with the Ubiquitous Anthropology project which we did with the Bororo population in Mato Grosso, in Brazil: they took what we designed to experiment on the next forms of anthropological writing, and used it for a really effective activist process in which they covered the governamental buildings in Brasilia with their political claims using augmented reality. Amazing!
    And wonderful opportunities there for information visualization (and interaction)
    ciao!

  3. JanWillem Tulp / Jun 15 2011 9:03 am

    Great post. I have to think about this a little further, but I wonder if the relatedness of the data being visualized is an important factor of whether a visualization should be more than a ‘wow-factor’. It touches somewhat on what Enrico Bertini said on the difference in process when you create a visualization for the public or for a small group of people who are actually solving real world problems, and they cannot solve that problem without visualizations (like in biology for instance).

    If you take Moritz Stefaner’s Notabilia, the data being (additions and deletions on Wikipedia) visualized is inherently not something that would influence my behavior, while the example you give (running) is a situation where the data is personally related to you, and second, you’re probably already motivated to improve your running performance. As it turns out, research on the influence of visualizations is still rather shallow and unknown territory, but perhaps motivation or willingness to be influenced may be a factor here.

    I do think that the way scientific visualizations are created is a goal oriented way, since real world problems are being solved. And I also think it is ok to have a different approach, the ego-driven and self-centered approach, if you’re creating a visualization that will be consumed by a wide audience (on the Internet for example).

    It’s a good subject for a discussion!

  4. Greg / Jun 15 2011 10:09 am

    Yes great discussion. There are many types of data visualizations, designed and developed by many different people for different goals. I personally like the work of ego-driven designers, but would never labelled them as such. It’s just people designing, thinking laterally and working through a process. And, engineers are just as much ego-driven when they talk about the ‘truth’ of their work🙂

    It’s similar to the one story vs 1000 stories that came up at Visualizing Europe discussion. It’s a recursive discussion, both work, both fail. It all depends on context & audience.

  5. erikduval / Jun 15 2011 10:37 am

    Thanks for the comments – and keep them coming😉

    Moritz mentioned on twitter: “measuring impact is zricky, but that should not stop us. Find some pointers here: http://fellinlovewithdata.com/reflections/can-visualization-influence-people-i-mean-can-we-prove-it (comments section)” and “also worth looking up: persuasive visualization”. Of course, persuasion is just one kind of goal that a visualization can have.

    Enrico’s point about essential visualizations reminds me of what Nicolas Negroponte sometimes says about research in Technology Enhanced Learning: if you have to measure the difference, it is too small… Things that really have an impact are beyond questioning. Or at least, you cannot question that they have impact. (You can still wonder whether that impact is only beneficial…)

    The one story versus 1000 stories argument is indeed related. My own take on that is that it is perfectly fine for a visualization to offer one clear story. But it would be nice if I can then become a more active user and change the story or explore alternative stories. And, echoing Gregor’s presentation, it would be even better if I can pursue this further and actually research whether the data are correct and complete or a subjective selection.

  6. Moritz Stefaner / Jun 15 2011 11:49 am

    Hi,

    good questions and thoughts, a few things:

    Enrico wrote two follow-up posts on his original one, which I should have linked to:
    http://fellinlovewithdata.com/reflections/data-visualization-influence-1 and http://fellinlovewithdata.com/reflections/data-visualization-and-influence-part-2-examples-and-links
    These posts provide some pointers to relevant studies, but IMO also show that the question of impact of visualization in general is maybe to broad to be answered in a satisfactory way.

    About the design methodologies: Allow me to bring in a comparison of Google and Apple, one of the two most successful tech companies today. Google is known for exhaustive user testing of every single design option, while Apple is known for “zero market research” and strong vision-driven design. Who of these two is “right” and does it “the proper way”?

    • erikduval / Jun 15 2011 10:39 pm

      Good points, Moritz. Of course, users often don’t know what they want. I often don’t know what I want. I sometimes recognize it when someone shows it though.

      Also, I do think that Apple does a lot of analysis on how users actually use the products and then learns from that. Then again, you mentioned you do too.

      I guess that there is no “right” or “proper” way. But they both know how to measure success, even if only after releasing the product. I worry sometimes that we are not so clear about what “success” means.

      And both Google and Apple accept that they fail trying. Think wave. Or mobileme. Do we know when we fail?

  7. Moritz Stefaner / Jun 15 2011 11:51 am

    read “..two of the most successful..”

  8. jerome cukier / Jun 15 2011 4:43 pm

    Hi Erik. I’m sorry we didn’t get to talk yesterday because I really share your views on what you call goal-driven visualizations.

    In my experience, many datavis practitioners overemphasize objectivity and truthfulness. This is to be expected from people with background in academia, journalism or government.

    There is another tradition of using data for advocacy (http://www.tacticaltech.org/sites/tacticaltech.org/files/infodesign.pdf). Yet, datavis is underused in politics and in lobbying where it could be put to great use. While this perspective is feared by some I am really looking forward to this and wording towards this with all my energy in France. As data will be an important argument in the public debate, it would lead to more rationality, more accountability, etc.

    I think interactivity is a key part of changing a person’s opinion. What set me in that direction is this talk of Jon McLoone from Wolfram (http://river-valley.tv/the-future-of-interactive-publishing/). if 30 mins is TL;DW the key passage is at the 6th minute. This relates to teaching, but that and convincing are essentially the same process of idea transmission.

    McLoone argues that teaching could be much more effective if instead of being based on static material it used the user input in a transparent way.

    instead of saying – this is a theory. here is an example. Accept that it works.
    you say: Here is a model. You can look the nuts and bolts if you feel like it. Now throw your data at it and see it in action.

    Now translate that to influencing opinion, with your smoking example for instance.

    Most of us believe that there is a link between smoking and health. Yet traditional communication is not too effective.
    Now imagine you have a slider with number of cigarettes smoked per day. As you move your slider you will see statistics about people with that tobacco consumption level (like life expectancy, or incidence of cancer, or lifetime cost of smoking, etc.). And yes, there must be a sample of data with that information somewhere.

    so as you move your slider you will see what happens if you increase or reduce *your* consumption.
    this gets the message across more effectively than a vague general message.

  9. erikduval / Jun 15 2011 10:46 pm

    I strongly agree with the value of engagement in learning, Jerome. My favorite example is how kids learn to ride a bike: you get very immediate feedback on how well you are learning. It hurts. You bleed. But you also directly experience your progress. And riding a bike is obviously empowering: you go where you steer the bike and as fast as your legs tell the bike to go. Most kids take the bleeding because they want the empowerment… I often think about how to organize my classes so that my students would be willing to bleed because they know they are empowered by what they learn.

    On the other hand, there is the question where ‘advocacy’ becomes ‘demagogy’. What if some Goebbels discovers the power of the visual and the active? Hopefully, openness would help to fight that kind of demagogy. Surely, the answer cannot be to keep the power away from us all because it can be put to bad use. But it still worries me sometimes…

  10. David Lloyd / Jun 20 2011 12:11 pm

    Erik

    Hi. You might like to have a look at Section 5.2 of my PhD at http://gicentre.org/papers/lloyd_evaluating_2009.pdf
    where I consider the design process for a visualization.

    David Lloyd

Trackbacks

  1. Goal oriented visualizations? (via Erik Duval’s Weblog) « Carpet Bomberz Inc.
  2. Erik Duval’s Weblog | Ra Puke Moana

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: