Correlation analysis is a vital tool in the hands of any Six Sigma team. As the Six Sigma team enters the analyze phase they have access to data from various variables. They now need to synthesize this data and ensure that they are able to find a conclusive relationship.
What is Correlation Analysis ?
One can best understand correlation analysis with the help of an example. Let us suppose that the management of a factory has come up with data which says that the as the shift time of the workers is increased, their productivity decreases.
However, as of now there is raw data and this is only an observation that some Six Sigma team member may have come up with after having a first look at the data. But Six Sigma methodology does not lie on opinions of the people involved but rather on objective fact. Correlation analysis will help statistically confirm the fact that this is indeed the case.
In order to perform the correlation analysis, there must be sufficient data for the variables under question. Once there is sufficient data, this data was plugged into a formula developed by Karl Pearson. This formula was famously called Karl Pearsons co-efficient of correlation. This involved complex calculation and mandated the presence of a statistician in the Six Sigma team.
However, fortunately nowadays most of the calculations are performed by a software tool. The humans involved must merely know how to add data to the tool and how to interpret the results.
Correlation analysis typically gives us a number result that lies between +1 and -1. The +ve or –ve sign denotes the direction of the correlation. The positive sign denotes direct correlation whereas the negative sign denotes inverse correlation.
Zero signifies no correlation. And the closer the number moves towards 1, the stronger the correlation is. Usually for the correlation to be considered significant, the correlation must be 0.5 or above in either direction.
Correlation analysis just confirms the fact that some given data moves in tandem. A dangerous implication that mangers make is of causality. Based on the correlation analysis it is impossible to say which variable is the cause and which is the effect? It is also likely that both the variables move in tandem because they are affected by some third common variable. However, these are just cases and the fact remains there are other analyses available to find out the causal relationship. However, in most cases the fact the variables have a correlation is enough to take relevant action.