iSixSigma

Deciding Significance Level

Six Sigma – iSixSigma Forums Old Forums General Deciding Significance Level

Viewing 8 posts - 1 through 8 (of 8 total)
  • Author
    Posts
  • #40186

    K
    Participant

    In any project (or in before taking up any problem), how to decide the Significance Level?

    0
    #123946

    “Ken”
    Participant

    P,
    Level of Significance is a parameter used to characterize the Type I error level when conducting a comparison study.  It’s used in statistical testing, not for setting up a project.
    In comparison testing, the level of significance establishes how conservative you want the test of hyposthesis to be:
    alpha = 1%     extremely conservativealpha = 5%     moderately conservativealpha = 10%   slightly conservative
    Way too much stats for the morning!
    Ken

    0
    #123947

    AB
    Participant

    I think we’re lost in the words again. From the question, it appears Priya is talking about a process for prioritizing process improvement opportunities.
    Priya – You would look at the organization’s balanced score card if one exists to determine which improvement aligns with strategic priorities and then assign some sort of a ranking to the improvement opportunity. This is the project/opportunity screening step of what is commonly referred in the project management world as  “Project Portfolio Management”.

    0
    #123956

    “Ken”
    Participant

    Well, that’s the other answer.  P’s capitalization of the words signifying a name may have push me in the stat direction.  Not sure which perspective he is truly interest in, but between us I think it’s covered. 
    If a Balanced Scorecard is not used how would you approach the answer?

    0
    #123963

    AB
    Participant

    You assumed P is a “he” :-). I would bet my money on the name being feminine.
    On question of what would one do without a balanced score card, I guess many of these tools serve as structured communication vehicles. Whether you call them Big Ys or balanced score cards or transformational maps, they all communicate.
    In organizations where communication isn’t happening as it ought to you would need to resort to getting in front of the right people to validate/justify which just makes process improvement that much harder.
    I am curious how would you approach it Ken?

    0
    #123975

    “Ken”
    Participant

    AB,
    Difficult to tell these days if Priya is a latin based name, or possibly Indian or Malasian.  If it is the former, then could still be male.  In most of these cultures, he, she, its, usually carries a similar meaning.  But, just in case, I extend my humble apology to Priya if I confused the gender.  I’m still curious whether the question was about project selection, or about statistical testing.  Hopefully, Priya will respond back to clarify.
    Concerning project identification and selection, I find many organizations do not have anything close to a Balanced Scorecard, let alone know of its use.  I’m not sure what your experience is in the strategic aspects.  Perhaps Mike could add his experience in this area when he returns to the states.  In these situations, I usually resort to using whatever seems to work with the culture of the organization I support. 
    As one recent example, I conducted a strategy session with the senior management of a large Pharma organization.  They felt they knew what projects they wanted to conduct, and the priority of those projects.  I thought this is great.  So, prior to starting the work I took them through an linear exercise where they identified the customer vs. product demand relationship-customer issues, concerns, and needs, and ordered the issues using demand data.  We next identified the processes connected to the issues, and the projects were identified and prioritized using customer demand and product info as weighting factors.
    In another organization, the management team had the projects identified by their own internal criteria.  They had 40 to a 100 projects, but didn’t know which ones to begin first.  They couldn’t do all of them simultaneously.  So, I sat with the management team and worked through a Benefits vs. Efforts Matrix.  The B vs. E Matrix is a simple Priortization Matrix similar to a C&E Matrix.  There were three dimensions I used for each of the two areas of the matrix, and I blended a Hoshin Planning approach to score each potential sub-area on a 1, 3, 9 scale for each of the projects.  Using this approach we  quickly, (with a day), priortized projects having the greatest impact to the organization, and identified the ones’s requiring the lowest level of effort.  We used Pareto’s law to narrow the focus to match the available resources. 
    Other approaches using risk management are applicable especially in the regulated industries.
    Let me know your thoughts, experiences, and ideas on project IDing and prioritization when you have a chance.  I’m always looking for new angles.
    Ken

    0
    #123982

    K
    Participant

    Thanks to both of you for your efforts.
    I still feel there is lot of subjectivity in deciding the significance level. In future also if you come across any formula to calculate it, pl. post it in the site.
    By the way, Priya is Female name in India.
     

    0
    #123987

    “Ken”
    Participant

    Priya,Ah, so you were referring to statistical significance. That being the case, we have a challenge. There is no formula for calculating it. It is selected per the criteria I suggested earlier. You might want to a look at a good old fashion statistics book to get some additional insight. Here are a few online def’ns:http://www.answers.com/topic/statistical-significancehttp://davidmlane.com/hyperstat/A72117.htmlThis one has some nice graphics:
    http://www.vias.org/tmdatanaleng/cc_level_signif.htmlGood luck in your search.Ken

    0
Viewing 8 posts - 1 through 8 (of 8 total)

The forum ‘General’ is closed to new topics and replies.