iSixSigma

Samples Infrences

Six Sigma – iSixSigma Forums General Forums General Samples Infrences

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #53736

    Lanning
    Member

    For materials that come into our plant (8 categories) we have, typically several supplier for each (2 to 10 suppliers). Of a concern is a count of an attribute, the count rate per piece received can vary widely, but the natural lower bond is zero. Currently the plant has the resources to sample one piece for each shipment (lot size varies from 50 to 500+).

    The issue that came up is how reliable is the current sampling plan? I did take two lots of material and had them, reluctantly; take 20 samples from the two lots. The lots, from 2 different categories of materials had very different modes. I know that the raw data was not normally distributed (a natural boundary exists, as stated), but also noted a few instances with very high counts (the highest is about 5x the average). I did try transforms and found a reasonable amount or normality introduced for a square-root transformation (the lower bound is not nearly the problem then).

    I have been asked to estimate the average of the other un-sampled lots and categories if only one sample is taken from each. I submit that with only one sample that I cannot infer anything about the mean from the lot, even if normality were demonstrated. Am I correct or just lacking education?

    0
    #191265

    Mikel
    Member

    Say what?

    Your question is as clear as mud.

    0
    #191266

    Lanning
    Member

    OK, perhaps I added too much informatio.

    A process is being examined, with the shape of the distribution reasonably known. Due to constrains, one is permitted to only draw one sample from a new lot. From that one sample, is it correct to say that no infrence about the mean/mode/average of the new lot can be made?

    0
Viewing 3 posts - 1 through 3 (of 3 total)

You must be logged in to reply to this topic.