Note: Images may be inserted into your messages by uploading a file attachment (see "Manage Attachments"). Even though it doesn't appear when previewed, the image will appear at the end of your message once it is posted.
Register Latest Topics

  Author   Comment  

Posts: 245
Reply with quote  #1 
With the recession being in the news, I was doing some web searches to get ideas for graphs to try, and I came across this BLS website:


There are lots of short articles here, and each one has a pretty decent (simple) graph.  The part that I was most excited about though, was the "Chart data - TXT" link below each graph! 

You can click there to get the actual data used in the graph (and not some fancy html or pdf table that you can't read into your own graphing software ... but rather a nice simple text file! :)

For example, the following article on "CPI":


Has a link under the graph, to the following data:


I think this is a really good idea for web-based graphs, and maybe even a candidate for a "best practice".


Posts: 1
Reply with quote  #2 
I agree Grasshopper...including links to data sources is indeed a good practice.

Unfortunately, the graphs themselves are not 'best practice' (note: Grasshopper never said they were best practice, and below I'm just pointing out where the graphs themselves fall short).

In my opinion, the example graph at  http://www.bls.gov/opub/ted/2009/mar/wk3/art04.htm has quite a bit of chart junk. Edward Tufte's descriptions of  what he termes 'chart junk' includes 'useless or optically active grids, boxes and frames around graphs, redundant representations of data (which I take to include redundant use of words), cross-hatched bars  - basically anything on a chart that that serves no inherent purpose.

I believe this chart could be made much cleaner, better etc by:
  • losing  the border around the plot area -  it's distracting
  • perhaps dropping 'FEB' from the horizontal axis labels, as it's already clear from the title that the graph data is from Feb to Feb.
  • getting rid of the dotted line gridlines altogether, or at least making them an unbroken (i.e. undotted) line, which is much less distracting from the 'meat' of the graph...the data series itself
  • Change the series to seasonally adjusted data, in line with the introductory paragraph 'On a seasonally adjusted basis, .....' Or make it much more clearer that the graph does not match the explanatory into paragraph. Granted, the overly long chart title states the graph is 'not seasonally adjusted', but its easy to miss this in this overly long title, and it's just weird that the purpose of the graph doesn't exactley match the purpose of it's associated words.
  • Change the time scale of the graph to a much shorter period - perhaps a year or two. The associated words that the graph is supposed to be illustrating  talks about a span of a year, and points out the difference between Jan and Feb 2009, yet the graph spans an incredible decade. In addition to the point above, this shows that this graph has very little to do with the words in which it is embedded. I'd personally plot 2 yearly series on the one graph, with the current year as a line graph and the previous year as a subtle area graph (bu subtle, I mean I'd use a very unobtrusive light background color, just so readers can contrast what happenned this year directly with what happenned the previous).
  • Get rid of either the 'percentage change' from the chart title or the vertical axis. It's redundant to have it twice. Perhaps this graph actually should be plotted using percentage points rather than percentage change, although I need to give this more thought when my morning coffee has kicked in.
  • Perhaps use a fatter series line, and definately use a different colour than garish green.
  • Get rid of the black vertical graph axis line. Tick marks and numbers are enough. 
Some readers may disagree with my opinion that a few of the above points will make the graph clearer. I'd respond that - in the very least - my suggestions would 'do no evil'...that is, they certainly wouldn't make the graph any harder to read.

I'd apply similar critique to the other graphs on the site I looked at. Too often the site uses graphs with textured  shiny finishes. The graphs I looked at on average don't do a great job of adding to the story that the site author is telling us via their words.

I'll whip up a redesign in excel (which is a  great charting engine when you get to know it intimately, and can bypass the defaults), and post it here when I get a chance. My above suggetions influenced by Edward Tufte and JohnPeltier, and am keenly awaiting delivery of books by Stephen Few and Mike Alexander so I can see what they have to say on the subject.

But reading Mr Few's content on this site, I don't think he'd disagree with much of what I've said above.




Posts: 245
Reply with quote  #3 

Looking forward your new/improved version of the graph!
Some changes sound good on paper, but don't work out well visually, so it will be important to actually "see" them, to know if I like them better or now :)

Previous Topic | Next Topic

Quick Navigation: