Graphical Summary (Data Visualisation)

Anytime I face an issue, I check the available data. It doesn′t necessarily lead to a project, but I have an understanding how the problem looks like. Additionally, I know that the collected information might have come from unreliable sources (inaccurate, MSA not checked, biased). So, drawing any conclusions may bring me on a false path.

What do I look for? Minitab has a superb tool, the Graphical Summary. I apply this tool to every data set. Although the name suggests that there are graphs (yes, they are), I look for the p-value. What is the significance of this value? It tells us whether the distribution is normal or not. If it is over 0.05, we conclude that the data have a normal distribution. If not, we may look for the reasons. Does any particular phenomenon drive the process? Are there outliers? Is there any sign of a measurement system problem? Possible stratification factors? Is there any long tail? Any sign of physical limits? Or any odd pattern?

I look at the graphs, the histogram and the box plot, but these are not so important, as statistical tests do not support them. Many may argue with my approach, but I am against data visualisation without statistical testing. ?Our brain has evolved to spot patterns. If we watch continuously changing clouds, we see different animals. But the human eye is not a computer, and sometimes we conclude and should not. Hence, we developed statistics to help us prevent these mistakes.

The Graphical Summary gives us the first impression. The second one is how it behaves in time. So, the next article is the Run-chart.

#projectmanagement #interim #sixsigma #projectleadership #medicaldevices #operationalexcellence