Not every process or problem will produce diagnostic data that we can use statistical or other mathematical tools to address. Sometimes we feel ill-equipped when we have a problem for which our neat and sophisticated tools won’t apply. What should we do?

We should remember that all our statistical, mathematical, and sophisticated modeling tools are based on basic fundamental problem-solving methods. The methods came first, and the math was developed as a means of applying the methods to numerical data. This means that even if we don’t have neatly organized numerical data, we can still apply the method. It just requires more logic and reasoning on our part, and less math.

Loss of process control is a common problem. When we have processes, typically chemical or machine processes, that rely on measurable or adjustable settings to produce measurable, quantifiable outputs, we can use control charts to monitor process control. Control charts track measurable variables and mathematically evaluate whether the results of each measurement are characteristic of normal process behavior, or if they are statistically uncharacteristic, meaning that something has changed or gone “out of control.”

 …