Comparing Sets of Data in Six Sigma

The goal within the Six Sigma Methodology is to find a way to manufacture a product or provide a service without any flaws, time and time again. Six Sigma means near perfection and the people who use this quality control measure have as their main mission to develop a process where the same function can be completed successfully 999,997 times out of 1,000,000.

Statistical data is analyzed by Six Sigma Green Belt and Black Belts in order to reach their conclusions.  This is precisely why these professionals and their colleagues must be very careful and diligent in both collecting and in comparing sets of data.  It is assumed that the sets of data which have been gathered have equal values but that homogeneity of variance does not always exist. There are often variables that need to be considered before a fair comparison can be made. This requires that an analysis of variance to be performed. The test is also known as ANOVA, for short, and will tell you if there are differences in the collected data that have to be factored into your equations.

Once the validity of the statistics is affirmed, Six Sigma experts will examine the data for variance in order to highlight areas which might need to be improved upon. There are two basic types of variance; one is known as common cause variance and the other is known as special cause variance. Common cause involves factors that are essentially out of your control as a business owner while special cause signals a flaw in the manufacturing process or service model.

An example of common cause variance would be the different grain patterns in pieces of wood. It is impossible to produce a product that looks exactly the same every time when you are working with wood because each piece has characteristics all its own. You (and your customers) have to accept that there will be some variation because of factors out of your control. What are unacceptable and within your control as a business owner are variances that occur because of special causes. In the example of a wood shop, a bad cut or poorly mitered angle would be an example of this type of variance which is correctable.

Six Sigma has worked over and over again to reduce variances in manufacturing and the service sector. The review is only as good as the data used to reach the conclusions about how to make the operation more efficient. Before crunching numbers, check for homogeneity of variance, or every calculation done from that point on could lead you in the wrong direction. The old expression garbage in, garbage out applies when doing statistical analysis within the Six Sigma Methodology.