Graph too clever by half

Mike Carniello writes:

I wondered what you make of this.

I pay for the NYT online and tablet – but not paper, so I don’t know how they’re representing this content in two dimensions.

I’ve paged through the thing a couple of times, not sure how useful it is – it seems like a series of figures – in a one-column, many-row display might have worked as well (or, perhaps, two-column (differentiating oil producer groups)).

I agree, I found it disconcerting that the axes start changing as I scroll down!

17 thoughts on “Graph too clever by half

  1. I find it confusing to see time going every which way–back, forth, up, down, and around. If they really want to get fancy with the graphics, they could go fully 3D. The z axis would be for time. I think that would make more sense than these graphs. Otherwise just use something simple like columns, as Mike Carniello suggests.

    • I’m not used to this kind of visualization either – the choice of axes, not the “guided tour” animation. Despite the labels, I keep trying to equate horizontal position with time, and having to backtrack mentally.

  2. I find the choice of graphic form (basically a scatter plot with lines linking the year-points) uncontroversial. Charts like this have existed for decades (http://www.econlib.org/library/Enc/PhillipsCurve.html)

    The changing scales confused me a bit at first, too. But then I thought that if you really want to focus readers’s attention on one region or country at a time —rather than letting them compare all countries to one another,— this may not be a bad idea. I took a second, careful look at the graph, and liked it much better.

    This could be a good project to discuss about in class: Should graphs/charts/data maps be understandable at a glance, or is it appropriate to make readers work a bit harder to decode it in order to gain some useful insights?

    Just my —preliminary— two cents.

    • I agree. I don’t think this chart is controversial (in fact I like it as a way to combine a time series with a scatterplot) and I’m not overly worried about the horizontal axis changing. If it didn’t change…we’d be complaining about that!

  3. I find the charts work well to illustrate the story narrated in the text, which explicitly mentions the axis changes (“When you shift the chart to show production by OPEC”, etc). But I agree that the animation is more flashing than illuminating. A series of slides with the three plots (OPEC, US and total oil production) statically displayed each time and using colours and labels to highlight the data relevant to the text in the slide would probably be clearer. But that would not be dataviz enough, I guess.

  4. Each individual plot is fine — price vs production (or vice versa) with a different dot for each time period — but I think they fell in love with the animation so they flopped back and forth between OPEC and World and US just because they could. I would rather see a 3×3 grid of plots: each row shows a different region, each column highlights a different time period.

  5. They should have had the plot with 4 panels under each other showing :
    x) Price of oil as a function of time
    x) World oil production as a function of time
    x) OPEC oil production as a function of time
    x) US oil production as a function of time
    That should have made things so much clearer…

    • Another approach would be to have a second (right-hand) y axis for price. That way, in each of the charts, you could show both price and production in relation to time.

      Or you could create a dynamic chart where the user could select any two variables to display in relation to each other. That would be pretty easy to pull off. There could still be a “guided tour” of sorts: a list of things to try, and an explanation of some of the phenomena that will appear.

      All in all, the current visualizations are too noisy–and I think the time plotting contributes to the noise, especially in certain charts. In the second chart especially, the dots are so densely clustered that you can’t always see what is going on.

  6. Perhaps not the easiest way to see all the information at once, but I find this a clever way to show a 2-D path over time. It is not a scatter plot, it is a path, as the order of the points matter. The scale of the x-axis matching the current view is disconcerting because you can still see the points of the remaining points transparent in the background, knowing that those points are on a totally different range of the x-axis. So there is room for improvement. I still think it is a good attempt.

    • The “transparent” points are also part of the chart. They are not on a different range of the x-axis (I don’t know that do you mean exactly). They are “hidden” because they correspond to a date outside the period of interest. For example, in the first OPEC chart the points before 2010 are highlighted and in the next one the points after 2010 are highlighted (and at the same time, the x-axis changes from OPEC production to World production so each point moves horizontally to its new position).

      • OK. I misinterpreted the graph. I “retract” my comment. I went quickly and thought they were doing something entirely different. Which clearly has to lower the opinion I have of the visualization and perhaps the opinion I have about myself.

    • Completely agreed, they deserve an A for effort, and certainly an A- at least in achieving their narrative goals.

      To the quibblers : Yes, there is some information overload in the weekly-scale plot. Also, the navigation dots are too small IMHO. But overall it’s an exciting example of modern, high-quality data visualization.

  7. Easiest to understand what they are trying to say if you use the row of buttons on the right to go up and down between the elements of the narrative. It’s confusing if you go from one panel to the next by mousing up and down on the diagram, but very easy if you just hit the button for the panel you want (and you can skip panels or go backwards easily).

  8. I don’t think this article was meant to viewed on laptops/desktops. I looked at it on my iPhone, flicking through the sections with my thumb. It works rather well in my opinion-you feel in control of the process and the small blocks of text moving through the plot break the visualisation into reasonably easy to digest sections.

  9. [Allowing that the following could be misguided and wrong… . (I believe that time is not a function of oil production.)]

    Thanks for listening!, if you do.

    I feel compelled to say that… it is sad that the NY Times can afford to pay for expensive (super-printable) graphing software… and does… but cares so little about graphing that it is *not even at the point of knowing* that it should employ someone minimally capable of visual communication. (… Not to mention that the operator of this expensive software ostensibly does not know how to use it (in the relevant sense) either.)

    Conversely, perhaps they know perfectly well how to graph oil supply versus oil price (it’s in the title!!), but prefer instead to impress people by making simple things complex (using expensive software).

    Please pardon me if I indulge myself.

    It is conspicuous that, whereas the article talks about price being a function of supply against demand… the graph *does not show demand* (although it does show price, which is basically (demand/supply).)

    Horizontal axis – time. (It is not a function of anything else, and it is linear (and it is relevant).)

    Supply is an independent input, not a function of something else; that is the point. Conversely, however, supply very much *is* overall a function of demand — just of other things as well.

    The article is about these other things. If a graph is indicated, its function is to show deviation of supply from demand; the article can then comment on this. The required graph thus simply shows supply versus demand over time. I would recommend one line showing (supply/demand) (as opposed to two lines — one each); it is not really relevant whether it is 1,000,000 vs 800,000 or 1000 vs 800, I suspect.

    I take it that the price simply follows {supply vs demand}. If so, it is merely a comment. If not, then… it is trivial to add price on a second axis (and, again, the article could comment on the deviations). (This would indicate inverting, to (demand/supply).)

    A 2D scatter plot, with time veering wildly over a 2D space, and with supply shown on both axes… [deep sigh].

    p.s. I do like the exploding detail, though. (Also, I do concede that it is useful to be able to show different oil amount scales (although conversely transitioning by entirely replacing the axis seems counter-productive).)

Leave a Reply to leoboiko Cancel reply

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