Agree with the other changes. With this level of analysis, who cares about the platforms with very small market share? If anyone is interested, a separate chart of “other,” broken down into individual platforms, can be created. Good work as usual.

The more I see the one on the left with the rearranged order of categories, the less comfortable I am with it.

I was more interested in the year-to-year trend, so I liked the line panels better, though both have their place. There was no missing 2008 data, by the way: someone confirmed that the spacing of the surveys was early 2006, late 2007, early 2009, so 18 months or so between data sets. I wouldn’t bother with dotted lines in the case of uneven intervals, unless I knew for sure there was missing survey data, or if the gap was substantially wider than the time between other surveys.

I’d guess that the platforms that had no charted data some years actually had around 1%. The data seems to have been cut off arbitrarily at the 9th platform. I would have tried to plot the platforms with the largest combined values, and plotted them in all years. If you don’t know what the small value is you could always apply “N/A” or a similar label. If one of the featured ones went belly up, I’d have used zero, if the overall total was enough for its inclusion. In fact, I would probably have shown only WordPress.org through Expression Engine, and included more into Other. More room for labels, less worrying about the missing values: were they around 1% or did they signify going out of business, because compared to the rest, who cares?

As to the question of which of the panel bars is the better option, I think that it depends on how the data was collected. For a one-time (static) chart with manual collection and arrangement of the data, I agree that the botton panel bar might be better. However, if the data is the result of a database query, then you can’t presume the list, and the first panel bar is the only option (you’d somehow have to alert the reader that the list of platforms aren’t the same from one year to the next).

One other thing. Suppose that one or more or the platform makers went out of business. For example, if the makers of TextPattern went out of business sometime in 2007, would it make sense to show this platform in 2007 and subsequent years?

Darren uses WordPress for ProBlogger. I thought he was one of the many users of the Thesis theme too, but I see a link to a designer in the footer of the page. I’ll be using Thesis, too, when I get a chance to implement it.

Good point about copyrighting and graphics having similar purposes, and how his graphical treatment here worked against his otherwise good writing.

I think that copy writing and data visualization share some basic principles, but when it comes to applying those principles something is lost in translation.

It would be interesting to link Darren’s charts to his own advices on writing and managing a blog, and show how these charts fail to apply those principles.

Can you guess what charting tool he’s using? Bloggers-about-blogging and internet marketers seem to love it, and I see that background all the time…

I prefer the second panel bar chart, which keeps the platforms in the same order. I preferred this one before I even drew the chart, because the mixed up order of platforms in the original pie charts caused me some initial confusion.. The platforms are sorted by total value in all four years in the panel line charts, because that’s how the Excel pivot table I used sorted them. The bar charts were done more manually, so they might use that order, or the order of platforms in 2009.

1. Order by size rather than alphabetically.
2. Use consistent ordering when comparing charts.

The top chart of Jon’s panel bar charts follows the first principle; the bottom follows the second. There are times when there is only room for one chart. My advice when two principles conflict with one another is to choose the one that is least likely to confuse the reader. Since some readers might not notice that the ordering has changed in the top set, I would choose the bottom one if forced to only use one of these two. The bottom one also highlights the fact that the share for some platforms such as movable type were quite different in 2006 and 7 from 2009.

These comments just apply to the case where the year is held fixed; I like also showing the line graph with the platform held fixed.