I need to present a chart (not really specified more than that right now) that will consist of two almost equal parts (ie ~50% each). However, a pie-chart would almost always look about the same, given that the diffrance remain small.

I need to find a way that illustrate and emphasize the larger of the two parts. A pie chart with 51%/49%-parts looks to dull.

One solution is to display a bar chart with a broken axis, but is there any other way? Accuracy (to the underlaying data) is not a great importance.

  • I am not sure this can be answered without knowing what you're really trying to show. Can you provide background, carefully disguised if necessary? What's important about the larger of the two parts? Why can't you concentrate on the 2% difference? – e100 Nov 30 '11 at 17:15
  • Well, I guess what I am trying to achieve is a way to emphasize the 2%-diffrence, in my mind that meant emphasize the larger of two parts... What would be a good way to illustrate the diffrence? – David W. Feb 14 '12 at 11:21
  • It all depends. Can you compare the 2% with other measures? Has it been less than 1% for five years? Is it greater than company X's 1.4% and company Y's 1.8%? Try not to directly compare the 49 and 51. – e100 Feb 14 '12 at 13:57

Create a pie chart and then separate the two pieces. Have them floating next to each other and throw on some effects (drop shadow, 3D). Use vivid, contrasting colors for the slices (blue and orange, purple and yellow). Put the percentages in big numbers.

|improve this answer|||||
  • Good idea, perhaps even enlargeing the greater pie-piece... – David W. Nov 29 '11 at 17:01
  • -1. None of this except showing the numbers will emphasise the larger piece. In fact, just choosing a pie chart is a sure way to hide differences between two similar quantities. – e100 Nov 30 '11 at 17:17

If you are looking to misrepresent, rather than present, you can just do what the big boys do and start the scale at or slightly below the lower of the two values, with the second value plus some small amount for the max value. Don't label the scale with any values and you are good to go.

Another way is to improperly use a logarithmic scale.

|improve this answer|||||
  • 2
    Nearly any technique can be used to 'misrepresent' as much as 'represent' from the standpoint of communicating a message to an audience. A scientist may not approve, but in terms of general communication, it's not necessarily good/bad right/wrong. – DA01 Nov 29 '11 at 15:53
  • Perhaps I have triggered some knee-jerk reaction in you, but the OP says "accuracy to the underlying data is not important." This is nearly the definition of misrepresentation. – horatio Nov 29 '11 at 16:56
  • 1
    Example. Let's say you need a product. Company X and company Y both sell it. X's product is rated at a quality of 51%. Y's is rated at 49%. They cost the same amount. Now a chart with an axis going from 0 to 100 would likely show the two as pretty much being equal and the CEO may simply flip a coin to make the decision. But if the axis is 45-55, it may become quite clear to the CEO that one is obviously more highly rated, even if marginally... – DA01 Nov 29 '11 at 22:12
  • 1
    ...One can certainly call that misrepresentation of the data, but I call that a proper presentation of the particular message that is trying to be communicated (one of the products is better than the other). – DA01 Nov 29 '11 at 22:12
  • 2
    And if the axis is not properly labeled, then the CEO thinks X is 20-30% better. In addition in reality the error between 49 and 51 % is larger than the difference between the numbers. This is a misrepresentation. – horatio Nov 30 '11 at 14:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.