Last time, I talked about the power of run charts and Pareto diagrams. (also see Hank Williams & the Seven QC Tools)
I've also found scatter charts to be an effective, intuitive tool for making connections between different metrics.
Pretty easy to use. Essentially, you develop a checksheet to gather data for a set of metrics you feel might be connected.
Then you plot the data, one set on the X axis, the other on the Y axis.
You can tell pretty quickly whether the metrics are related.
Of course, correlation does not mean causation.
Because the sun rises when the rooster crows does not mean the rooster caused it!
We must also have a plausible mechanism. An hypothesis, if you will: Why does this metric affect that one?
Here's an example. Some years ago a client company was facing serious safety problems.
They were a heavy equipment manufacturer whose motto was "Safety First".
Their safety problems corroded their relationship with their team members.
We analyzed a dozen critical metrics; we spent days on the factory floors trying to grasp the situation.
The combination of reflection and action was powerful, as always.
We used scatter diagrams to identify the key driver of safety performance - worker involvement.
There was a strong inverse relationship between safety incidents and involvement - and there was a plausible mechanism.
Our hypothesis: If I believe this is my process, my machine, my plant, and if I have the ability to change things for the better, safety incident rates will go down.
We later proved the correlation with a more sophisticated ANOVA analysis, but scatter diagrams got us there.
Best regards,
Pascal
No comments:
Post a Comment