Every day we are flooded with data, be it stock index, billing figures, scrap rates or cricket scores. We also see the there is variation in all the data we receive, but the capability to understand and handle variation is very important.
I can't think of any process which doesn't exhibit any sort of variation, or even if we don't see any, it would be our incapability to measure the variation. Managers who are presented with data need to make decisions using data and the variation in data from day to day, month to month or quarter to quarter. One may have come across managers who appreciate and award team members when there is a month's dip in scrap level or a rise in billing figures; at the same time get tough when they see a dip in trend the following month.
Managers use data trends not just to make business decisions but also to award or even fire people
What is important however is to know when to act on variation and when to leave the process alone. If the data from a system is plotted as a graph and control limits plotted, then trend of data points with respect to the centre and the control limits will give an indication if the process is stable with time and in control. Anybody with basic knowledge of statistics or the ability to use any statistical software can plot a graph with control limits. What is important is not how to calculate and plot the control limits, but to understand that some variation is acceptable and some are not.
A process is stable and in control only when it is under the influence of Common Causes; and a process which is out of control is said to exhibit Special Cause Variation. So every process does exhibit variation but one needs to act only if the variation is caused by special causes; else action will be mere tampering with the process which is also called as over adjustment.
Common cause (also called Noise) is present in every process and is produced by the process itself; and to remove or lessen noise a fundamental change in process is required. On the other hand, Special Cause (also called Signals) is unpredictable and typically large in comparison with common cause. Special Cause is caused by unique disturbances or a series of them and can be removed or lessened by basic process control or monitoring.
To understand this conceptually, let us look at an example. Consider an office bus which is scheduled to reach office every day at 8 am. The bus in its normal course picks up staff at various points in the city and arrives at office around 8 am daily, not exactly at 8 am, but hovers around an average of 7.55 am. Let us say the arrival could vary from say 7.50 am to 8 am, so we see an overall variation of about 10 minutes in this case. On a given day the arrival varies from the average of 7.55 by 0 to 5 minutes in either direction and there can be no assignable reason for this variation. Under this situation the process is said to be under the influence of common or chance cause variation. If the vehicle has a flat tyre on a particular day surely there is an assignable or special cause and the vehicle could come late by say about 30 minutes; and in another instance if there is a city wide strike and all schools and government establishments are closed, the bus could reach by say, 7.40, again due to a special cause.
This knowledge of variation and common and special cause is very important for mangers to take the right decision and not to mislead by variation under in stable process, and also sporadic brilliance at times. This is applicable for all process including cricket scores, and surely the Indian team scores will be hovering around an average, though could vary depending on the game being played on the home ground or outside, and also depending on the opponent. But for sure there is an average and a variation and there can be some sporadic incidents which I call magic or misery depending on the case. To make a shift from the average either there has to be a continuous improvement or a breakthrough. For breakthroughs there has to be a proper analysis of the process and of the root causes of problems, to enable breakthrough improvements.
To summarize I would say understand variation, understand the central value or average, don't be perturbed by shifts in a controlled variation; but plan for breakthroughs to make a significant change and also to reduce variation.
Pradeep Kumar E.T. A Master Black Belt in Six Sigma , is the Country Manager- Operational Excellence with Tyco Electronics Corporation India Pvt Ltd. Feedback can be e- mailed to firstname.lastname@example.org
Issue BG73 Apr07