iSixSigma

Difference between Segmentation and Stratification

Six Sigma – iSixSigma Forums Old Forums General Difference between Segmentation and Stratification

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #33242

    Pallab
    Participant

    Hi,
     There are few confusions exist around DMAIC life cycle. When in Measure phase during data collection we brainstorm on probable segmentation factors and collect data for all of them. After that we plot data to display process dependency on segmentation factors. Are we not here trying to come out with the root cause in the initial phase of DMAIC ?
    But, in analyse phase, we again stratify data if the means of two sub groups(based on segmentation factors) are different. Here what we do , do we need plot and find out PCB for each of the sub groups and run improve phase for both the processes.
    Please guide me.
    Pallab.
     

    0
    #89628

    DaveG
    Participant

    Pallab,
    Before I comment further, what is a PCB?
    [email protected]

    0
    #89631

    Allegood
    Member

    PCB is printed circuit board?
    Did you see this old thread on Stratification versus Segmentation?https://www.isixsigma.com/forum/showmessage.asp?messageID=28494
    I still don’t have a really simple answer to the question either. I’d be interested in learning from someone else.
    Thanks,Troy

    0
    #89648

    Pallab
    Participant

    PCB : Process Capability Baselining.
     

    0
    #102415

    Arora
    Participant

    Hi Pallab,
    Looks like a very old post but lets start it again.Are you saying that the root cause analysis starts right in the “Measure” phase and not just in “ANALYSE” because we brainstorm the segmentation factors and find each fragments impact on each other or on the defect of our interest?.
    I think, YES. but, I would call that “Logical cause analysis” and not “Root cause analysis”.And as you indicated, if PCB has to be done, then we should also run the SPC to check for the special causes if any(I still remember the good post on this subject by you for my question),and if present, they ahve to be eliminated before PCB.I guess though the role of each phase is defined, sometimes, the phases do share it’s roles between them.
    Just a thought…
    Thanks,
    Kris.

    0
    #103122

    Kris the IInd
    Participant

    Hi,
    As you have rightly said Root Cause Analysis starts in the Measure phase itself because here we are defining segmentation factors, or in other terms we are groping for some X that may affect our Y.
    But the basic question still remains, what is the exact difference between segmentation and stratification?
    This is my understanding of the same, and would request someone to correct me if I am wrong.
    Segmentation by external factors helps us find the drivers of variation. In the loans example, which was stated earlier in this thread, categorization by size ie. small, medium and large is segmentation.
    Stratification involves ‘layering’ of the process, and studying the variation drivers in each layers, ie. each process step or a group of process steps. In the above example, for stratification we will see what is the delay induced in the loan application stage, in the credit stage, etc..
    That is why when comparing two processes,
    if you have the same mean and different variances you can proceed to RCA
    if you have different mean and the same variances you segment and then stratify each process. Because then you can find the causes of variation specific to each segment and then compare processes and try to equate the means ( if required ).
    if you have different mean and different variances then you stratify processes and then define segments. Because that will give u both the layer and the segment which is driving the variation.
    Regards
    Kris the IInd

    0
    #103141

    Ron
    Member

    Let’s start with basic definitions:
    First in the Measure phase of a DMAIC process you are gathering information regarding your problem statement (No RCCA here).
    When taking random samples you can perform this task in many ways
    A stratifed samples is performed by first determining what the samples size is and then using the population distribution to detemine how many of each subgroup you will sample. For example in Age we may strify to 10% (20-25) 20%(26-36) 40%(37-53) 30%(54-above)  If I sampled 100 people I’d take 10 from grom 1 20 from group 2 etc.
    Segmented sampling is similar but on a larger scale such as male, female, or community by community, ethnicity etc.
     

    0
Viewing 7 posts - 1 through 7 (of 7 total)

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