3 thoughts on “Chart Wars: The Political Power of Data Visualization

  1. Some impressions;

    (+) The narrative is lucid and concise
    (-) The graphic illustrating efficacy of visual presentation over aural is, embarrassingly, a poor visual presentation, and the point is better made aurally.
    (-) The graphical literacy examples are unwonderful. I've no idea what the "axis and origin" example is meant to show. The "omission" graph omits a huge swath of presidents in its 'improved' version. For "causation vs correlation", there's nothing causal illustrated. And saying "pie charts suck" without exception is just wrong.
    (=) Perhaps it's necessary with the audience and the format, but the jokes to content ratio seemed a bit high for me – compare it with e.g. the first 5 mins of Hans Rosling's Ted talk. Why not take that time to explain something where a novel data visualization really helps?
    (=) The slides on 'Chart Wars' and the headshot on "Pictorial Superiority Effect" seem unhelpful. But maybe they allow the speaker a bit of flexibility in this (slightly bizarre) presentation format.

  2. George – thanks for all the feedback. Allow me to address your comments:

    – thanks for the praise, I was heartened to see you found something good to say about it.
    – yes, quite ironically, I found the slide illustrating the efficacy of visual presentation over aural to be the most difficult to create. I look forward to seeing your superior alternative…just remember that you can only do it with one slide, little to no animation, and the audience will only see it for 15 seconds.
    – my "unwonderful" examples: the axis and origin slide was animated, and sadly the video didn't capture it – you'll have to trust me that with the illustrations the slide made sense. The 2nd omission graph was not meant to be an improvement, but rather an alternative (and again, space constraints were a paramount concern). The causal relationship implied is from a fairly well established (tongue-in-cheek) internet meme. See http://venganza.org for the details. On the pie charts: I agree. And indeed I use pie charts from time to time. But for the purposes of this presentation and this audience I had to simplify the message.
    – Yes, the format definitely matters here. It is a happy hour type atmosphere of young tech/biz/entrepreneur type people, and there is plenty of alcohol involved. Informality is a must. The other key thing I would point out is that you refer to the "first" five minutes of Rosling's talk, indicating he had more time following that for his talk. I had only five minutes. That's it.
    – And finally, again, due to the format, you definitely need some breather slides in your deck.

  3. Thanks Alex

    On visual/aural efficacy; you're just illustrating two numbers (75% vs 10% information retained). Two bars with labels would do it… maybe (horror!) even two pie charts.

    On omission; I think it would work better to keep the graph essentially the same, but extend the x-axis to the left; a "reveal", if you like.

    On causality; the problem is that a scatterplot, on its own, doesn't state anything about presence or absence of a causal relationship. So I don't see the relevance of the example to graphical literacy.

    On timing; in 5 mins, Hans Rosling describes the change in global fertility and life expectancy over 40 years (and gets in a couple good relevant jokes, at the start). He could have stopped right there, and it would still be a good talk.

    (++) Huge credit for standing up in front of a happy hour crowd and engaging them with anything remotely statistical.

Comments are closed.