Stratification is used mostly in the define stage of a six sigma project. When to use the stratification technique is still on the decision of the people involved in the project. However fairly good cues are given by the fact that the spread of the data is too large. If the points are scattered all over the curve then it is probable that the sample being taken is wrong.

Stratification is basically a technique to solve the sampling error. It is one of the seven basic tools of quality used in six sigma. Here is how to use stratification for better results.

Results are too General ?

A good sample ensures that most of the points lie within close proximity of the mean. This is not always the case, but attempts should be made to ensure that this stage is reached. This is because meaningful conclusions are likely to be made in this stage only.

Hence if the Six Sigma project team observes the following phenomena in their projects they must consider using Stratification:

• Results are too broad or general to draw any specific conclusion
• There are a large number of exceptions to the statistical conclusions that are being drawn

The stratification process can be used either while the sampling is done, if the project team suspects that the results will be too broad or they can be implemented after the results have been obtained and found to be unhelpful.

Go on to Sub Process Level

In case of large processes with many steps, stratification normally needs to be used. Thus the Six Sigma team must understand this problem and use a layered approach. This means that the processes can be divided into a number of sub-processes. These sub-processes can then be further divided into smaller segments. This process can go on until the last activity has been reached. However, the project team must know the data it is trying to collect and stop at the point where they are likely to get what they want from the process.

Divide Data into Homogenous Strata

While sub-dividing process using stratification, it is not necessary that the sequence be followed. In most cases the data being sought is pertaining to the underlying causes. Therefore if the same operator or the same machine, facility etc are being used at different stages, they can be subgrouped.

The idea is to make the data as homogenous as possible without making the data collection exercise unmanageable or too complex.

Draw More Specific Results

The process is iterative. Results should be drawn to understand whether they are actionable. If not, further stratification should be employed.

Stratification is a basic but vital tool of quality management. The results of the process largely depend upon the data collection plans involved and hence they must be carefully designed to meet the needs of the organization.