iSixSigma

Box-Cox Transformation – Normal Non-nomal data

Six Sigma – iSixSigma Forums Old Forums General Box-Cox Transformation – Normal Non-nomal data

Viewing 7 posts - 1 through 7 (of 7 total)
  • Author
    Posts
  • #51190

    Joseph
    Participant

    Hello All,
    I am comparing a normal (before) and non-normal data (after). The non-normal data was transformed into normal via box-cox.
    How do I compare the two data…original data vs. transformed data?
     
    Joseph

    0
    #177056

    Mikel
    Member

    No, All data has to be transformed in the same way.Transforming is unnecessary. Go read the white paper on SAS’s
    website about the irrational fear of non-normality.

    0
    #177058

    Joseph
    Participant

    Hello Stan,
    Could you reply with the link to the white paper. Thanks.
    By the I need to perform a 2 sample t test or Anova on these data.
    Joseph

    0
    #177062

    Ken Feldman
    Participant

    Arggggg.  You shouldn’t be playing with that tool unless you know what you are doing.  Here are the steps to take with non normal data:
    1.  Understand your process better and be sure that there is not some specific reason for the non normality such as poor stratification or combining two groups that should be analyzed separately.
    2.  Depending on how non normal you are, both the t test and ANOVA have some robustness to a situation of non normality.
    3.  Feel free to use non parametric testing such as Mann Whitney or Moods Median
    4.  If all else fails then consider transforming your data but make sure you know what you are doing.

    0
    #177068

    Joseph
    Participant

    Awesome! Thanks Darth.
    Joseph

    0
    #177069

    Chris Butterworth
    Participant
    #177070

    Mikel
    Member

    Chris,Thanks, I didn’t see Joseph’s question until now.

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

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