iSixSigma

CI in MTB

This topic contains 3 replies, has 4 voices, and was last updated by  thevillageidiot 13 years, 8 months ago.

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #42465

    annon
    Participant

    How do you use the CI in MTB hypothesis testing (ANOVA, t-test, etc) in comparison?  I am getting p-values indicating true differences but my CI overlap….what am I missing?  Thanks!

    0
    #134053

    Darth
    Participant

    It shouldn’t although the CI graphs are not always really accurate to scale.  Can you post an example?

    0
    #134063

    BTDT
    Participant

    Annon:Keep in mind that the 95% CIs in the graph are for each individual subgroup and not the overlap itself. Imagine you have a 95% CI on two groups. When the two CIs just touch each other, the probability is 0.05/2 * 0.05/2 = 0.00063. It is possible that the two regions will overlap, but still will have a probability of less than 0.05.ANOVA will calculate the same CI for each subgroup using the pooled standard deviation, while the t-test can calculate CIs for each subgroup (if you uncheck the “assume equal variances” button).Cheers, BTDT

    0
    #134080

    thevillageidiot
    Member

    Thanks guys…very helpful.

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

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