thank you for your comment and implied question. You are right: it appears that better communication and coordination would have made this episode less frustrating to everyone concerned.

What makes these events also applicable to Six Sigma is the Catch-22 between “lack of data” and “defining the problem.” In the Define phase of DMAIC our challenge is to understand and agree on the parameters that allow us to quantify the problem and measure progress. When we deal with routine situations, we have industry norms and best practices to guide the problem definition. Things get interesting when knowledgeable people disagree about what the norms and best practices should be. In this case, some people felt the need to be conservative in the name of safety, while others preferred to assess the risk in less stringent terms.

A successful Six Sigma project leader needs to be able to harmonize conflicting needs and priorities in order to complete the “Define” phase. Often data can help, but there are times when stakeholders disagree on what the data means and how it should influence our actions. To me the Eyjafjallajokull eruption is a case study of how difficult “defining the issue” can be.

Not sure why this was linked under six sigma or how six sigma methodology could have made much of a difference (though better communication may have), but you’ve provided some good links.

So true.

The question here shouldn’t be whether the methodology failed or the people failed. It should be whether the methodology was implemented, explained, and managed properly by the people who decided to put it in place.

Six sigma is a great methodology, it’s not magic. It still requires work, management and checks/balances.

Thank you for your thoughts and reading recommendation – Senger’s book looks very interesting and I can see why you recommend it in this context. Defining a problem goes far beyond technical specifications and process steps – it also has to address the “human factor” and it appears that Senger’s book addresses that quite nicely.

As I have since learned, L6S synthesizes ideas from many quality models and it appears that quite a few concepts derive from Deming’s PDCA model. One nice thing about PDCA is the fact that iteration is more obviously built into the model and doesn’t have to be implied like in Six Sigma’s DMAIC approach. Maybe some folks would argue with me about this, but I think with Six Sigma there is a temptation to approach everything as a straight-line project with a clear beginning and end. At least that is my impression after hearing people’s stories in class.

I agree process improvement techniques like Six Sigma have direct application for improving data collection, analysis and distribution processes. In addition, quaity improvement techniques provide a useful tool set for improving data quality.

I use the PLAN-DO-CHECK-ACT model. In addition, I recommend The 5th Discipline Fieldbook by Peter Senge & others. Senge’s book dives deeply into mental models, systems thinking, team learning,and the balancing of advocacy and inquiry. Highly recommended.

Every hierarchy (business, government, for profit, non-profit) exists as system for delivering value to customers. Systems are made up of processes – processes are composed of tasks.

Gut-feeling and experienced insight can sometimes be exactly wrong. Tools like Six Sigma provide fact-based insight.

I look forward to future posts on this blog regarding how you have employed the six sigma toolset on your own projects. Please share those.

I enjoy reading your blog. Keep up the good work!

my sincere apologies for offending you and for omitting the link to your post. Neither a personal attack nor a critique of your skills and experience were intended. Your post described how you explored this data and mentioned that the scatter plots were not the best way. I added the “why,” included the comment that “… the graphic below appeared in a recent blog post …” and forgot to add the appropriate link.

By the way, my reasoning for using a bar chart rather than a trendline: the underlying data was a collection of responses to separate surveys. Since we have to assume that different people responded to each survey, I did not want to imply that the opinion of the same group of people has changed over time. Maybe over a virtual beer we can discuss whether I am splitting hairs and whether the samples were representative of the population

Just a question, were you suggesting that I do not have the appropriate experience and skills? It sure comes off that way.