Data Viz Done Right

August 29, 2016

Makeover Monday: U.S. Companies Are Hoarding Trillions Offshore

9 comments

First, I must give credit for finding this week’s Makeover Monday to Jade Le Van. She’s a massive supporter of the work everyone has been doing in this project. Thank you Jade for finding this beauty from CNBC.


There are so many problems with this visualisation. However, the most egregious mistake is that the data is completely wrong. The author took the original data set, which has the total money held offshore by a company and allocated that same amount to every country that each country has an affiliate in. For example, General Electric has $199B stored in 18 offshore subsidiaries spread across seven countries. The author allocated $199B to EACH of those seven countries. That’s simply wrong!

What doesn’t work?

  • The data is inaccurate.
  • The pie charts are a ridiculous choice, especially when you show every single company in each country.
  • The companies aren’t sorted.
  • The size of the pies isn’t proportionate across the difference columns. In other words, they aren’t all based on the same range.
  • It’s impossible to find a company you might be interested in investigating.


What works well?

  • Nothing; the whole visualisation is a complete disaster.


I had a few minutes to explore this data set while on the bus home, so I opened up the data in Vizable. Within about 20 seconds I had two views that were far superior to the original.


Quickly and easily, I have a view of just how much money companies are keeping offshore and who they are in a nice, simple ranked list. Great! I already have something much better than the original.

Two more swipes in Vizable and I’m now looking at the companies with the most offshore subsidiaries.


At this point, I felt I basically had all I needed to tell the story in Tableau. I quickly reproduced the two charts from Vizable, then added a Pareto chart for more context. Lastly, I added some headers above each chart to tell a synchronous story.

9 comments :

  1. I like the simplicity.
    And even thought there may be other stories to tell with this data, this at least tells one!

    It's unfathomable that any news organization actually published the original.

    ReplyDelete
  2. Great viz! I'd love to see a Tableau Tip Tuesday on that Pareto Chart... the dual reference lines, percentile and dynamic tooltips and title are very elegant. The in/out color scheme is great too...

    ReplyDelete
    Replies
    1. Sure no problem. I can do it this week.

      Delete
    2. Here you go... http://www.vizwiz.com/2016/08/tableau-tip-tuesday-how-to-create-two.html

      Delete
    3. On that note, I am curious if there is any specific reason for the 65% and/or 30 company break point, or was it just a matter of a point that looked good?

      I ask only because it always seems most fitting to me, on a Pareto chart, to note the ~80% mark, and see what percentage of the group accounts for it - though obviously this is not any sort of mandate for Pareto charts.

      Not a critique of your choice, just curious if there is any particular thought behind it.

      Delete
    4. I played around with different numbers and felt 30 told the best story. I actually have this set up as a parameter but decided not to show it.

      Delete
  3. Hi all, see my take on the makeover, love to hear your feedback: http://mfitz0610.wixsite.com/mikefitz/us-company-foreign-tax-havens/2016/05/03/How-to-navigate-the-full-color-palette

    ReplyDelete