Responding to Incapable Processes - Six Sigma


When a process is stable but incapable we need to understand the common causes of variation in the process in order to decide what action to take to improve the overall variation. The approach and simple tools discussed earlier will be suicient in many cases to achieve acceptable levels of improvement. If this is not the case, or we wish to have a more robust answer, there are several more sophisticated approaches we can take:

  1. Establish whether the process distribution is different under different conditions: This can be done using tools such as Hypothesis testing, ANOVA or Non - Parametric tests.
  2. Understand whether variation in a dependent variable can be explained by variation in another variable: This is usually done by means of scatter diagrams, correlation plots and regression analysis.
  3. Test a variety of conditions to establish which factors contribute significantly to variation: This is the area of designed experiments.

Design of Experiments

An experimental design sets out to investigate whether a series of factors, when varied have an effect on the variable of interest (usually referred to as the ‘Response Variable’ or ‘Quality Characteristic’). It is a more structured way of approaching the kind of ad hoc experimentation which goes on in a lot of organizations. All experimentation follows the same basic approach:

  1. Deane the Experimental Goals: We need to clarify what we are looking to achieve from the experiment and the scope of the investigation.

  2. Select Response Variable (Quality Characteristic): This will usually be the key performance measure of the process; the thing we are interested in optimising; the ‘Critical Y’ in Six Sigma terminology.

  3. Choose factors, levels and ranges: Brainstorm / Cause and Effect analysis can establish potential factors which may affect the Quality Characteristic. Factors can be continuous (e.g. how much milk we add to our tea) or discrete (e.g. do we add the milk before or after the tea). We will need to change each factor at least once to observe the difference it makes. Accordingly we shall select 2 (or more) levels for the factor, ensuring that the range is suicient to have an effect, but not so large as to move outside reasonable ranges.

  4. Select Experimental design: Given the question you wish to answer, the number of factors and levels and resources required we can select an appropriate experimental regime.

  5. Perform the Experiment: Ensuring that experimental error is kept to a minimum.

  6. Analyse the outcomes: Using appropriate techniques for the design chosen.

  7. Draw conclusions and make recommendations: Taking care to test the recommendations to ensure that the experiment has not been compromised in a way which we did not spot.

There are a number of approaches to experimental design, and it is not my intention to compare them in this text.

All rights reserved © 2018 Wisdom IT Services India Pvt. Ltd Protection Status

Six Sigma Topics