I had a chance encounter with this chart recently - and whilst I "get" what the school was trying to show (attendance matters and all that), the chart left me perplexed.

Firstly, the overriding convention with axes is that as you travel further away from the origin, the value on that axis increases. In this case, as I travel left to right, I expect % Attendance to be increasing. So, I would conclude that "the higher the attendance, the lower the A*-C attainment. But, that's not what the bars are labelled. In fact, the exact opposite.

Secondly, the chart is suspiciously regular - and each bar is "just the right size" to show a gradual change as attendance varies. Looking at each bar, we see that the widths, or data ranges for each one is different (93.5 to 100 = 6.5, but 92.5 to 93.4 = 0.9) - For this to be a histogram, the width of each bar needs to reflect the data range. For it to be a classic bar chart, we'd expect the data range for each bar to be the same -- this fails both tests.

Thirdly, "Chances of success at GCSE" - is about as vague as it gets. I am presuming that this is in some way an aggregated measure across all subjects at KS4 - but nowhere on the chart (or on the web site) clarifies this. What about "double award" GCSE subjects - how are they factored it? What about BTEC and other vocational qualifications?

Fourthly, gridlines and data labels with an unnecessary precision (74.3% - really?)

Fifthly, and this might just be "picky" but I really don't like a chart making a "judgement" on the "success" at GCSE - charts should just present the data - the reader decides if it's a success.

Same data, but with some Excel-fu:

Call to action:
  1. Stick to graph conventions (left to right = increase)
  2. Make a histogram a histogram and a bar chart a bar chart (or learn how to do it)
  3. No gridlines
  4. No "judgements" on a chart - let the reader decide