Tag Archive: market research

  1. Define “Better”…

    1 Comment

    In the visual realm, what people “like” is by definition subjective.

    What “works” and what doesn’t in the context of dataviz design, however, is usually something one can determine a little more easily.  Effective data visualization is a combination of what’s visually appealing and what best communicates the creator’s ideas.

    In a 2-part series this week and next, we’re going to take a look at some visualization examples and question what “works” best in terms of being effective and being accurate.  As we’ll show, sometimes it’s a simple matter, and sometimes it’s not.  In most cases, the nature of the data shown will drive the most effective way of displaying it.

    Example 1:
    These charts show the same data points but in very different ways.  Line or trend charts are usually most effective in showing data trends over time; hence the name “trend”.  Showing data across non-date/time categories doesn’t really lend itself to using a line chart.  A simple side-by-side comparison of figures is really suited to using bar (or column) charts.  Hence, Choice B is more effective.

    Example 2:
    Pie charts are designed to show what share of the “pie” each category has.  Using the same data points, the bar chart is again the better choice in this example because it’s a simple numeric comparison.

    Example 3:
    This example is not quite as straightforward and perhaps a little more subjective.  But unless the column colors in Choice B are tied to their respective categories, there’s no reason to use different colors, except just because you can.  For that reason, using a single color for the columns looks less “busy”.

    Example 4:

    If we stick with the “single color” idea, then Choice A would probably be the better of the two, unless you’re an especially big fan of electric lemon yellow, or that’s the client’s corporate color.  In which case, we feel your pain because we’ve been there.

    Example 5:
    Even without the prevailing consensus in the wider design world that “flatter is better”, we think the simpler design of Choice A is nicer to look at.  Just because you CAN contort a chart into an angled, cylindrical format…

    Example 6:

    Should be fairly obvious…  we hope…

    Example 7:

    Enough with the zeroes!

    Example 8:

    This last comparison may not register so quickly.  Go on, take a few seconds, we’ll wait…

    *checks watch impatiently*

    You have to look carefully, which is the inherent danger of using a non-standard Y-axis range on a chart.  (Did you catch it unprompted?)  Unless Choice B appears in the context of other charts which also don’t start their Y-axes at zero, the user’s expectations are that it would start at zero.

    At first glance, the charted column values in Choice B are quite big (c’mon, Shirts, pull your weight!).  But checking the actual numeric values will show the gaps are not so dramatic after all.  Of course, doing a chart like Choice B CAN be used to influence the viewer to a certain point of view, should that be the goal of the chart designer.

    In our next posting, we’ll take a look at some examples that are a bit more definitively “wrong”…


  2. 3D – Good for HD, Bad for DV

    1 Comment

    The past few years, the world has gone mad for 3D.

    Practically every movie that comes out now is available in 3D, featuring the obligatory scene where it looks like a piece of something-or-other from an explosion is headed straight at you from out of the screen.  3D HDTVs have been available for a few years now, allowing you to wear the giant idiotic-looking 3D goggles in the privacy of your own home (although that particular 3D wave seems thankfully to be receding without having made much impression).  Sometimes, film 3D actually works quite well and appears seamlessly and goes nearly unnoticed when it’s done perfectly.  Most often, however, it feels forced and was done by producers just because they could.

    While 3D works at a degree in the world of entertainment, when your primary objective is visualisation and communication of BI or market research data, it by and large doesn’t work.

    The proliferation of easy-to-use dataviz software on the market means that more and more people are able to produce and publish their own charts.  That’s great, but it ultimately means there will be many examples of not just “bad” charts, but charts that are totally misleading in what they present.  One aspect of this which is a particular bête noire in the dataviz community is the 3D chart, and in particular, the 3D pie chart.

    People typically create 3D charts simply because they can, and believe (often mistakenly) that they’re more “sexy” than “flat” charts.  We can recall, during our formative years in MR, using a great charting package called Harvard Graphics.  It was so cool and revolutionary when it came out, and yes, we too produced 3D charts (*we collectively hang our heads in shame*).  However in today’s world the communication of information through data has become increasingly mission-critical, and so we need to ensure that the message we are conveying is utterly accurate.

    Let’s look at an especially egregious example….


    At first blush, the headline stories that we may take away from this chart are

    • The majority of sales took place in the East (I mean, look at the SIZE of that wedge!)
    • Sales in the East were roughly triple those in the West (bad job, West!)

    However, if we display the actual value labels to this chart we see a different story…..


    Now we can see that

    • Most sales were in the North (well done, North!)
    • Sales in the East were only double those in the West (still, c’mon, West!)

    So you can see how just by attempting to sex up the data by adding 3D perspective to the chart, we’ve totally changed the message the data conveys.  In fact, if it were a normal 2D pie-chart, the correct message would be conveyed without ambiguity as shown below


    The takeaway should be to keep it simple, because extraneous “noise” like 3D can easily distort the data’s message.

Shopping Cart