Six Sigma summarizes its problem statement as a mathematical equation in the form of Y=f(X1,X2,...,Xn). The Y here represents the Dependent, Output or the Effect and X is the independent, Input or Cause.

 

In order to improve the outcome Y, which could be either shifting the mean or reducing the variation in the process, the systematic approach is to control your X's which are your causes.

 

Now the question comes will all X's make an impact on Y. Let's look at an example.

 

I am a 2 Wheeler manufacturing company and see my Sales are gradually declining quarter on quarter. I get my Quality Management Team to review the situation and come up with an action plan. The QM team, post completing the define and measure phase, initiate the analyze phase and using techniques like brainstorming sessions gather all the causes for the decline in Sales. Listed below are the pointers
1. Increased Government Tax structure having bearing on my overall price.
2. Poor promotional offers vs. Competition
3. No Proper Customer Service / Demos
4. Increased rains and bad weather forecast in major cities making customers to speculate
5. Poor feedback on Post Sales Support.
6. Design Issues - Not appealing to  youth

 

Now let us categorize each of the above inputs into 3 categories,
a. Noise Inputs (N) -  Inputs(X) that impact output variable (Y) but are difficult or impossible to control. In our case we can keep the Increased Government Prices and Bad Weather forecast under this bucket. As a company, I don't have any say on both these things.

 

b. Controllable Inputs (C) -  X's that can be changed to see the effect on Y's. Also know as Knob variables, these inputs has the ability to change the output within the process setup. In our case, Design Issue is a Controllable Input. However, this needs some time as it has to go back to design team, researched, developed and implemented.

 

c. Critical Inputs (X) - X's that have been statistically shown to have a major impact on the output Y. Also, in our Pareto it can be on top call driver list where the likelihood of occurrence is most likely and controlling the same would be Somewhat Easy to control.

 

We can now see how the overall Cause has systematically broken down into Noise, Controllable and Critical Cause and my Critical Cause can be termed as Root Cause.

There are multiple high-quality responses to this question that have indicated deep understanding, experience and maturity in Root Cause Analysis. 

The chosen best answer is that of Sujata Dhawase - Three perspectives for root causes and a clear example explaining the perspectives. 

 

For readers, there is a wealth of knowledge on root cause in many responses, some of the notable ones are mentioned below.  

 

- "Interaction between causes" by Steve C

- "Contributing vs. Root Cause" by Steve C

- "Minimum and sufficient causes that eliminate the problem" view by Kevin

- "Economic Elimination" view by Kevin

- Crisp definition by Stephen J

- Detailed definition by Leanne 

- "Multiple causal chains" view by Paul

- "Scope of control" view by Chris-

- "Influential factors" approach by R Rajesh

- "Controlled Experiments" approach by Ransingh

- "CNX Classification" by Prashanth Datta

- "Primary root cause" concept by Chitra

- "Actionability requirement" by Parsa

-  "Contributing cause" by Girish Vasan

- "Practical Significance" view by Padmanabhan