iSixSigma

RCA – Multiple Root Causes?

Six Sigma – iSixSigma Forums General Forums Methodology RCA – Multiple Root Causes?

Viewing 12 posts - 1 through 12 (of 12 total)
  • Author
    Posts
  • #54759

    Padmanshu Kumar
    Participant

    Hello All,

    I was performing the root cause analysis using 5WHY which yielded 2 causes. I know this is purely based on our practice and experience, however what should be my next step or tool to identify which is the primary root case and which is the secondary root cause? I identified the primary and secondary root cause however few of my team mates have the opposite views (Per them, the secondary root cause is the primary and vice-versa). I am attaching the file which will show the 5WHYS, please assist.

    In general as well, when there are more than 1 cause identifies during RCA, what should be the methodology to identify which one is the primary root cause, should that be FOR-AGAINST?

    0
    #197005

    Prabhu V
    Participant

    Hi,
    After studying your case, I would like to mention the following:-
    a) The case study clearly highlights about the communication gap from central team to other stake holders.
    b) Production team can be right on their action since the practice till the moment was considering the lastest one as approved one.
    c) In order to avoid the same in future central team should give the clear communication to all stake holders especially on approvals.
    Just my opinion.
    Regards
    Prabhu V.

    0
    #197006

    Chris Seider
    Participant

    Follow the DMAIC process and numerically confirm the proportion of each root cause’s impact on the problem.

    0
    #197007

    Chris Seider
    Participant

    Follow DMAIC and numerically find how much each cause contributes to the problem. If they are so confident on the reason, data should be readily available.

    0
    #197020

    Padmanshu Kumar
    Participant

    Dear Chris/Dear Prabhu,

    Thank you so much for your valuable inputs.

    0
    #197022

    PRASHANT TIRAKANNAVAR
    Participant

    Evaluating using FMEA works best when analysis end up with more than one cause identifies during RCA.

    0
    #197027

    MBBinWI
    Participant

    @padmanshu don’t listen to @prashanthst – FMEA is a prevention tool. You can use a cause/effect matrix to identify strengths of effect and importance.

    0
    #197043

    Nik
    Participant

    So your main question had to deal with how do you identify the primary and secondary root cause? Ask yourself, why is picking one to be the primary important? Why not solve both?

    My experience has been that incidents are rarely the result of a singular event or failure. Often (especially when there is a singular incident), there is a “perfect storm” of several causes coming together to create an ideal environment for the failure to occur. This is one of the problems with the “5-Whys”; if you aren’t careful it leads to a very linear thought process and it takes a skilled facilitator to prevent the team from developing tunnel vision.

    Another item I’ve learned is that the Six Sigma and Lean tools have their limits on the types of problems they can effectively solve. For example, if it is a repeated problem relative to a process, the tools and DMAIC method are ideal. You can control chart it, gather data and using statistics determine the effect of various X variables on your problem (the Y variable), or do some process mapping and determine where the problem points in the process are. However, if it was a one-time accident, a singular incident, there may not be the data to chart or analyze. There may be no process they were following. So the Six Sigma tools, the Lean tools, and even the DMAIC methodology may not be ideal. I know there are some (my colleagues included) who think DMAIC is highly universal and can be applied to everything – – but there are better tools for investigating singular events.

    For investigating singular events, Ishikawa diagrams and the 5-Whys can be helpful. A few years ago I came across Cause Maps. ThinkReliability.com does a good job at explaining them. I’ve found this mapping technique not only effective at going deep, but also broad and mapping that “perfect storm.” You can actually see the root structure of the problem (using a plant metaphor). It makes it easy to see the contributing branches and shows you that if you don’t sever each one, the problem (the plant) could come back.

    So, I would suggest not limiting yourself by classifying primary and secondary causes. If you’ve narrowed it to 2, then take them both out. If you are investigating singular events (where there isn’t clear data, a process, or a pattern of events) then check out Cause Maps.

    0
    #197064

    Russell
    Participant

    I agree with Nik, it really does not matter which of the two is primary root cause, fix them both. If you had 10 or 20 then you would want to use C&E matrix along with any supporting data to support or deny stated root cause hypothesis’s.
    Regards,
    Russ

    0
    #197140

    Padmanshu Kumar
    Participant

    Great Explanation Nik, Thanks a lot for your valuable suggestion and knowledge sharing, it is really helpful.

    Thanks Russell…

    0
    #197214

    Shelby Jarvis
    Participant

    Another important consideration. 5Y and other similar techniques are actually identifying potential root causes. They are extremely effective. However, they are only as good as the inputs.

    My thoughts for what they are worth.
    1) Test your potential root causes. The true root cause(s) will turn the problem off and on.
    2) If you find both are real, then the advice by the early replies to statistically test becomes even more important. You can priortitize to prevent the greatest risk.
    3) If you find that one of the potential root causes has no clear effect, you can use this knowledge to dispell that portion of tribal knowledge and potentially remove controls currently related.

    I hope this helps.

    0
    #197218

    Mike Carnell
    Participant

    @Shelby Jarvis From your response I can see you are an advocate of Occam’s Razor which is a principle of parsimony, economy, or succinctness used in problem-solving devised by William of Ockham (c. 1287–1347). Basically the concept of “Keep It Simple Stupid.”

    That was probably a good idea in the 13th and 14th century. It is currently not the 13th or 14th century and hasn’t been for quite a while. You are essentially proposing brainstorming which is really what 5Y’s are. Turning it on and off – good idea but it really helps to understand what and why it is turning it on and off.

    Doing this type of weak hearted problem solving is how companies have gotten to the point where it is one band aid on top of another. Remember Einstein’s quote “We cannot solve our problems with the same thinking we used when we created them.” So in your proposal (particularly the 5Y) where is the new knowledge coming from so that we don’t have the “same thinking” that created them.

    This whole concept that there is someone walking around with the answer to every problem we face and just won’t tell anyone until someone specifically asks them is a completely inane concept. It is unbelievable what people will buy into to avoid actually have to do some work to fix a problem.

    To the original post – multiple root causes in not unusual but a single root cause is. next time you run a DOE and you find a single factor that accounts for 100% of the variation let me know. The whole concept of R-square is so you can judge how much of the variation is accounted for in your model.

    Just my opinion.

    0
Viewing 12 posts - 1 through 12 (of 12 total)

You must be logged in to reply to this topic.