Perhaps you could compare the chart I linked to in my previous comment with this one:
http://bizviz.jorgecamoes.com/mil-dados-mil-graficos-parte-5/en/
It is much easier with this one to understand the whole dataset if we just skip the “proportion of a whole” thing.

Tushar: “it is the internal mobility of the image which characterizes modern graphics. A graphic is no longer ‘drawn’ once and for all; it is ‘constructed’ and reconstructed (manipulated) until all the relationships which lie within it have been perceived.” Jacques Bertin wrote this 40 years ago, and you can imagine (or see in his books) how difficult was to accomplish this dynamic manipulation of the data with the tools of the time. Are we better now? We have better tools, but I am not sure. An average user still can’t do it in Excel. Also, “dynamic charting” should also be in our mindset, and in general is not. Two problems that must be solved.

The debate on what kind of chart is “best” has been framed by many (most? all?) as though the be-all and end-all is the printed document and a static display of data.

You know what? We miss out on a wealth of opportunity by accepting these ‘rules.’ And, it is especially true in this discussion since there isn’t even a mention of a printed document!

A little reorganization of the data and with the correct software, one can get a very powerful and dynamic analysis tool. Once done, an overview of the distribution of the national debt was a click or two away. Various US entities owned 75% of the debt. Of the foreign owners, East Asian nations (as I defined East Asia) owned just over 50% of the debt. The next largest region was Europe at about 25%. Of the European countries, Britain owned over 1/2 and Germany was a distant second.

Some years back I wrote a prototype add-in to support a PivotChart drilldown capability. Unfortunately, it’s still in that nearly same prototype state. {grin} Of course, it works with a pie or a bar PivotChart and as it happens it works with 2007. If you try the add-in, try both types of charts and you’ll quickly see why a bar (or column) chart is so much better than a pie chart. (*)

I reorganized the available data by adding two columns (Major and Geog), ending up with 4 columns altogether: Major (US/Foreign), Geog (for US:USA, for Foreign:geographic regions), Owner and Amount.

The XLS and XLSX files are at: http://www.box.net/shared/8ucs5j4xj7 and http://www.box.net/shared/clz7vxhv2o

The drilldown add-in is at
PivotChart Drill-down
http://www.tushar-mehta.com/excel/software/pivotchart_drilldown/index.html

(*) Do keep in mind the add-in is a prototype. Don’t ask too much of it. It works just fine with the data files above (and the PTs and PCs in them) but who knows what will happen if you try something…shall we say “more sophisticated?”

“When I decided on the single pie chart, I had the typical journalist in mind.”

So did you use crayons, or finger paints? But seriously….

What pie charts are best at is showing relative portions of a whole, and they’re really not so great at that. The same sized wedge viewed at a different orientation or filled with a different color may appear different, especially if the author didn’t have the good sense to keep the chart in 2D. Once you get beyond about five wedges, the pie chart is no good.

I’ve been following this debate with some amusement, and decided to add my voice to the din

Cluttered Pie Chart

- If you want your readers/audience to quickly spot the relative importance of a bunch of data, show a pie.

- If you want your readers/audience to be able to find out the exact relative importance of a bunch of data: use a table.

A table simply takes more time to study before the message comes through. But is more exact.

The problem is, a pie chart is better than a bar char to show proportions of a whole, and that’s fine. But if you limit the data points to 5, probably you should use a small table. Its usefulness is limited by the number of data points you can plot.

There are hundreds of examples of how useless a pie chart can be. I also have an example, if you want to take a look: http://bizviz.jorgecamoes.com/mil-dados-mil-graficos-parte-2/en/ (sorry about the automatic translation).

But what if the goal is to create a chart that requires only a *single* mental step for a person to see what portion of the total federal debt is owned by China, instead of *two* mental steps?

When I decided on the single pie chart, I had the typical journalist in mind.