iSixSigma

Tests of Normal Distribution

Six Sigma – iSixSigma Forums Old Forums General Tests of Normal Distribution

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #30640

    boettler
    Member

    Would be interested in learning typical procedures others use in checking for normal distribution? I have set of data with several variables. The Shapiro-Wilks Test indicates significance for several, but graphical plots look decent. Thanks.

    0
    #80042

    Cone
    Participant

    Go with the fraphics, the assumptions of normality at very robust.

    0
    #80043

    Cone
    Participant

    Sorry that is go with the graphics.

    0
    #80044

    Ashman
    Member

    The first thing I always ask myself is “do I have to have normality?”  There are a lot of things that can be done without normality.  If you must have normality, than I would say that the graphics would be fine as well.  Some of those tests are based on assumptions that may not be true for your data.  The only other question to ask is “is the significance important?”  Sometimes people get so caught up in the significance level that they forget that what their testing isn’t worth the money or effort it would take to make it normal.  I only bring that up since it sounds like you are borderline normal/not-normal.  If your p-value is in the 0.04 to 0.05 range, but the graphics look okay, then it probably isn’t worth the added cost to make the procedures normal, and then start improvements.

    0
Viewing 4 posts - 1 through 4 (of 4 total)

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