How to get the most out of your graphical analysis

The Six Sigma toolkit includes a large range of graphs, from Histograms to Pareto Charts, and from Box plots to Scatter plots. They can all be found in Minitab, and can often be created in a click of a button. So, what’s the problem?!

Well, the challenges tend to fall into two categories:

1) Deciding which graph to use
2) Understanding what the graph is telling you

1) Deciding which graph to use…

Describing what you want the graph to do will often help you to find the right graph. You might find yourself saying things like:

  • “I want to look at the distribution of my process lead times”
  • “I want to compare Supplier A against Supplier B”
  • “I want to understand the process performance over time
  • “I want to see if there’s a relationship between the input and the output”

The keywords in the sentences are highlighted, and these are the words that will help you select the right graph, as follows:

  • If you’re looking at distributions, then histograms are often useful.
  • If you’re comparing groups of data, then try box plots.
  • If you want to see trends over time, then a time series plot or run chart might help.
  • If you think there’s a relationship between two factors, then a scatter plot could be useful.

As with everything, it takes a little practice. You might also find the Data Door Routemap on page 114 of our pocket guide useful.

The most important thing is to decide which graph you need BEFORE opening Minitab! (or other similar software). In other words, you should be telling the software which graph you want, not the other way round!

2) Understanding what the graph is telling you…

Once you’ve created a graph, the challenge is to understand what it is telling you. The messages contained within a graph might not be immediately obvious, so the key is to take your time and focus on using practical language when interpreting a graph.

It’s also useful to remind yourself what you might expect to conclude from a particular graph, for example:

  • When looking at a histogram, you’re trying to understand the shape of the distribution. Is it smooth or does it have any unexpected peaks/troughs? Is it symmetrical or skewed?
  • When looking at a box plot, you’re looking for differences. Do the sub-groups have different ranges or medians, or longer tails (‘whiskers’)?
  • When looking at a time series plot, you’re looking for trends over time. Is the process increasing or decreasing? Does it have any obvious outliers? Would you describe it as consistent (quite stable) or very variable (unstable)? If the process changes over time, how quickly does it change – minute by minute, day by day or month to month?
  • When looking at a scatter plot, you’re looking for relationships. Do you think that the two factors are correlated? How tightly clustered are the results? Are there any data points that don’t fit the correlation?

And finally, our top tips for interpreting graphs…

  • Keep your conclusions practical.
  • Don’t read too much into a graph.
  • Don’t expect too much from a graph! Often, there will be only one or two key conclusions.
  • Remember how much data is in your graph (sample size) when drawing conclusions.
2019-05-10T11:11:09+00:001st April, 2019|Categories: Article|Tags: , |