Last week MSN released a report entitled Blogging in Britain (PDF via The Daily Telegraph) which is all very interesting and supplies lots of hard and fast numbers which I’m using for a work project. But whoever worked on the presentation of the stats in the report really needs to go back to school and learn how to accurately reflect charts.
Throughout the document, there are some nice looking 3D pie-charts. Except that pie charts should almost always represent the individual parts of some whole; all the bits added together should represent 100% of the whole with no duplication within segments (e.g. you’re only in one age-group). Yet the pie charts throughout this report show entirely different things.
By way of an example, here’s the first chart in the report:

At first glance, it looks as though the largest part of the blogging population is under 25s followed by 25-34s. And this is the case, but a close reading of the key shows that there’s something wrong with this chart. The percentages add up to 119%. Even allowing for the odd rounding error, you shouldn’t manage that with 4 data values. What they actually mean is that 42% of under 25s blog, and 58% don’t. You could show that as a pie chart, but this one is very misleading. The data should really be presented a little like this:

(To further see how the pie chart is wrong, consider the case where, say, 95% of the population blog, and it’s constant across all age groups. You’d end up with a pie chart divided into four neat equal segments looking for all the world like 25% each, yet each representing, somehow, 95% of their individual population segments).
This may all seem a little pointless, but it’s an indication of mathematical illiteracy at large, and a strong mathematical education is vital to strong debate.
And yes, I did spot the typo – “25-43” in the original chart.
One final note, in an interesting analysis piece on the Telegraph’s blog, Shane Richmond compares this survey with a seemingly similar one produced by advertising agency Universal McCann back in May. The two surveys have some startling differences that he does well to explain away. But he also mentions in passing that the MSN sample size was 743 internet users. Nowhere on the PDF is this explicitly stated. I assume he either contacted MSN or was working from a slightly more detailed press release. Once again, if MSN are going to put together a glossy PDF of their findings, a little detail about the methodology should really also be included.