DSA Insights

Instead of calling them pie charts, we’ll just call them donut (doughnut) charts and make them magically delicious.  Wrong.  These two charts, which appeared in the July 14th issue of BusinessWeek, are some of the worst charts I’ve seen in a while.  Note: in the print version, the legend doesn’t cover up the actual chart.  I’m not sure why this wasn’t fixed in the online graphic. 

There have been many articles and blog posts written on the ineffectiveness of pie charts, so I won’t belabor the point more.  For more information about why pie charts (donut charts with the center filled in) are not a good data visualization option, see the links below.  There’s no point in saying the same thing in a different way.  The timing of this post is good being that Seth Godin recently created a lot of stir about bar charts, which led to more pie chart discussions.

Related books:

Creating More Effective Graphs by Naomi B. Robbins

Show Me the Numbers: Designing Tables and Graphs to Enlighten by Stephen Few

Related links on discussion of Bar and Pie charts:

DSA Insights

Jon Peltier

Peltier Technical Services

Jorge Camoes - Charts

Juice Analytics Original

Juice Analytics

Stephen Few

Edward Tufte

Junk Charts

Note: the first pie chart is credited to William Playfair roughly 200 years ago.

I just can’t seem to help myself when it comes to sharing beauties like the pie chart below with my readers.  The pie chart was found here.  Yes, it’s colorful.  Yes, it’s an example of pimping your chart.  And finally, yes, it’s extremely ineffective.  Just recreating this chart accurately was nearly impossible because of the small slices that are not even visible.

It takes entirely too long to go from legend to chart, legend to chart, legend to chart, (you get the point) to cross reference the two.  A simple bar/column chart would have been a much better choice.

Here is my version that took me longer to figure out which values referenced which slice percentages than creating the actual chart.

Suggested Reading:

The Visual Display of Quantitative Information, 2nd edition

Visualizing Data

Show Me the Numbers: Designing Tables and Graphs to Enlighten

Creating More Effective Graphs

A few months ago, I wrote about an alternative way to present fractions (parts of a whole) instead of using a typical and flawed pie chart.  The first graphic below comes from BusinessWeek and the second one is DSA’s Excel created version.

Jon Peltier (Microsoft MVP) of Peltier Technical Services and PTS Blog, recently wrote me to tell me about a workbook in Excel, which contains VBA code to create something similar to our dollar visualization.  The results, which only took me a few minutes using Jon’s workbook, are below in the third graphic.  It’s a fairly useful tool to do something different for your audience.

Finally, the last graphic was created from a similar workbook in Excel by Andy Pope, Microsoft MVP.  The workbook (link below) contains instructions regarding how Andy was able to accomplish splitting the image.

In the related section at the end of this post are links to the workbooks and web sites of the creators.  Feel free to check them out!  If you are interested in how I created the DSA Insights version, feel free to contact me and I can walk you through it.

Thank you Jon and Andy!

BusinessWeek Version:

DSA Insights Excel Version:

Jon Peltier Excel Version with Easy Macro:

Andy Pope Excel Version (different Picture) with instructions:

Original Post:

Spending InfoVis, DSA Insights

Resources:

Jon’s Workbook, Jon Peltier

PTS Blog, Jon Peltier

Version of splitting a graphic, Andy Pope

Andy’s Workbook, Andy Pope

The Back of the Napkin: Solving Problems and Selling Ideas with Pictures at Amazon.com

The old stack ‘em up trick…

This graph is renders beautifully. From the x-axis labels being shortened to the lack of chartjunk in the form of gridlines and excessive tick marks. The colors definitely make the chart stand out. I don’t think many would argue that point.

If you take a look at an earlier post, I featured another visualization taken from the same page of BusinessWeek’s January 28th issue. As you will see, the color scheme for the whole page was green and yellow.

This graph shows two points very well!

1. The first point is the trend of “The Detroit Three” shown in green below. I can quickly and accurately see the initial jump, plateau and then gradual decline.
2. The second point that is definitive is the total for both the Foreign series and The Detroit Three series. The total sharply rises until 2000 and then starts a gradual decline over the next seven years.

What is almost impossible to determine is the change in Foreign sales from year to year and over the entire sixteen years. Intuitively, I can see that after about year 2000 the yellow portion is larger because the total is about the same while the green section gets smaller. Simply put, the reason a stacked bar chart is a poor choice when comparing more than one series over time is due to the baseline not being the same for the second (yellow) series.

For this example, the point isn’t what the total sales were over sixteen years. The point is how one compare to the other. I know this from the title of the page being, Detroit is still behind, despite hard-won gains.

Point: When showing more than one series over time (time being the key here) the most logical choice should be a line graph. I’ve seen some horrid stacked bar charts with many more segments. On a rare occasion, I may use a stacked bar chart only when time or trending is not a factor.

In the December 10, 2007 copy of Businessweek, I came across the horizontal bar chart below. Fundamentally, it depicts employee versus employer annual percent change in health insurance costs from 2004 to 2007. As you can quickly see, the horizontal bar chart is lacking in effectiveness unless you tilt your head ninety degrees to your right. So, I added a few of my own that really only took a few minutes to create.

Businessweek version:

Here is my Excel version using approximate values.

Here is my Xcelsius version with approximate values.

What I found interesting is that when you adjust the size of the graph, you get a more dramatic slope in both lines between 2006 and 2007. I wouldn’t recommend changing the size to maximize your theory or objective. Also, I do not like that the y-axis starts at 2% instead of 0% in the Xcelsius version. (Recommended reading: How to Lie With Statistics)

Below is my version using Open Office , which is an open-source (free) project that contains most of the products commonly found in the Microsoft Office Suite.

Finally, below is my version using Google Docs, which had minimal formatting options that I could find. I’m not even going to post the Many-Eyes version due to its lack of formatting. If you want to see it click here .

My preference is either the Excel or Xcelsius version, which illustrate the 4-year trend much better than the horizontal bar chart. Both took about the same amount of time to fine-tune. If I was limited on budget and didn’t want to spend any money, I would go with Open Office over Google Docs. There are many more formatting options in Open Office that help to create an effective data visualization. The only drawback is the amount of manipulation it takes to get from the default graph to the ones shown above. Both Excel and Xcelsius default to a horrid looking graph that I wouldn’t recommend using (both shown below).

Default Excel Chart:

Default Xcelsius Chart:

Which one(s) do you like the best? Would you suggest another option not found here?

In the fall issue of USAA’s magazine, I came across the chart below that shows fatal crashes, per 100 million miles driven and per age group.  I found it interesting, yet logical, that sixteen year-olds and those aged eighty-five plus, have the most fatal crashes. 

What concerned me about this chart is how distracting the non data-ink appears.  Here are the top issues I see with this chart:

1. The picture in the background.  It contains a picture representing each of the different age groups.  The graphic is not needed to convey the message properly.
2. The horizontal grid lines in the bar chart (chartjunk).  They are definitely distracting and take away from the data points.  Plus, there are no quantifiers on the Y axis to make them semi-useful. 
3. The yellow bar is difficult to see.
4. The cars are not needed at the top of the bars.
5. The data is not really insightful.  The only thing I can quickly understand is that the first and last category are a lot more than the middle one.

One hundred million miles sounds like a huge sample size, right?  Not really!  If the average driver travels 15,000 miles a year, then that’s only tracking roughly 6,700 drivers per age group over a year.  I would have expected the sample size to be much larger to make the study more statistically significant.

As software gets more sophisticated, charts and graphs seem to take on a whole different look and feel.  However, the one constant is the data.  Products such as Crystal Xcelsius can reduce the effectiveness of data visualization if used improperly.  It becomes more about how “pretty” we can make a bar chart or gauge look versus simplicity.  I’m sure you’ve heard the adage, “…lipstick on a pig”.  As a businessperson, I want to quickly and accurately understand what a data series is telling me. 

Alternatively, I also understand that many executives and the average person is intrigued and sometimes in awe of these “tasty” data representations.  That’s where balance and a true analytics expert can bridge the gap between what is simple, yet appealing and effective.

PART II - September 27, 2007