My What a Lovely, Yet Worthless, Picture
Thursday, October 14th, 2010 | Author: lisaksimone

I recently joined the Visual Analytics group on LinkedIn as I drag my many ad-hoc and homemade visualization techniques into the 21st century.  Representing data without bias, but with clarity and with usefulness is hard.

Years ago, one of my favorite data mining profs intoned, “If tortured enough, the data will confess.” - Lyle Unger, UPENN.  Whether good or bad depends purely on intent.

One discussion in LinkedIn focused on an odd graphic from the Wall Street Journal article “Broader U-6 Jobless Rate up to 17.1%: Why the Jump?”

What the heck does this picture mean?!?   It’s got some red stuff on it, should I be worried?

The poster commented and asked for feedback,

The WSJ is using a treemap of unemployment rates which seem to me less effective at communicating changes in unemployment rates than a simple line graph.

Now me, being all rabid about user interfaces and the like,  found this graphic to be beyond useless.  Not only did I get frustrated trying to figure it out, but I blew off the entire article as a result.

Before I say a word … go check it out.  Is it a simple bar chart?  A line chart?  Is it in any way recognizable as something you should have experience deciphering without batting an eye?  If not, then it behooves the author to describe, in sufficient detail, how to interpret the chart.

OR, the author should reconsider charts that need detailed explanations.  Can a straightforward explanation overcome the barrier to short-attention-span understanding?  Or more bluntly, is the chart there for information, or for prettiness and art’s sake?

We often get these ideas mixed up.

Holy crap that’s a cool pic, but what the heck does it mean?

While posting anyone elses’ comments here is inappropriate, here is my contribution to the conversation.  Awaiting your thoughts…

I’m no visualization expert but human interfaces and human factors issues bug me a lot. That being said, this chart baffled me. First, I didn’t realize it was interactive. Sure, it’s the internet, but give readers a clue. After pondering, I ended up concluding years progressed top to bottom, left to right.

Okay, next colors gave some clustering cues, but with no axes definitions, context was pretty useless. Even the article title mentions 17.1% but the chart colors range from 2 to 11 (percent?)

Then I noted the “Show recessions” white block. Again, I misread that as “Show(S) recessions” and incorrectly assumed the white/whiter blocks were times of recession. It’s not clear the white block is a check box, when the plot itself is full of white blocks (ok, rectangles). At that point, I gave up on the whole article.

Only after reading your responses, I see how easily it can be misinterpreted. There’s more info in there than I originally saw, but readers shouldn’t have to work so hard.

At a minimum, I’d expect to see an explanatory sentence or two directly below the table to translate how to read this non-obvious presentation, and I put my $0.02 in for the axis labels too.

I tried to be nicer on LinkedIn, but over here I’ll be honest - if this were over at ICHC-land, it’d be on FailBlog.

PS - is this a Real Life Debugging Problem?  YOU BET!  If an author/designer/writer/company is unable to convey their message, failure of one sort or another awaits.  (You just hope this chart isn’t your current tire pressure before that long ride home to NJ-land tomorrow.  ‘Cause that little red tire on the dashboard is one of those symptom-thingies.)