But boxplots in Excel are a thankless task – unless Microsoft has finally built them in.

I ran a similar grunt recently following an interview I had with a French newspaper – where the graph I built for a given example was discarded as it was not “traditional ones” (ie: line, histograms or pie chart…). People generally never stop and ask themselves “Do I really need a graph? If Yes, which one ?”.

In my opinion, the trouble we are facing is that people are trapped in a particular mindset regarding graphing data. This mindset is the consequence of the following elements:
- Constant exposure to a limited library of graphics: low quality of defaults-charting options in whatever system they are using + mass production of those “standard graphs” in day-to-day media exposure (Press, TV, Internet, as well as enterprise-based presentations and reports)
- Simplicity of creating a graph that shortcuts the necessary data-preparation work (data quality, data order, choice of graph based on pre-defined communication objectives or analysis results)
- Near zero level of awareness of the importance of “When and Why” use graphs (and how to build efficient ones) during educational / academic years before graduation.

What I found particularly sad is that people are generally not aware of the precious time lost in trying to figure out a bad graph. Imagine at a Company Board level, 12 CxOs individuals loosing 5mn each trying to figure out what is the reality behind glossy charts… That’s a whole hour lost and when you compute in their hourly rate… well the maths will not likely prove that this glossy chart yields on strong return on investment…

That time could be better spent in actually making decisions and implementing them. As you stated, endemic chart overload is no good contribution to progress.. And to counter this will require long-term evangelization efforts…

My advice would be – if you have to build a pie chart, make sure its a good pie chart.
http://www.cockeyed.com/inside/nuts/nuts.html