iSixSigma

Regression with Indicator

Six Sigma – iSixSigma Forums General Forums Tools & Templates Regression with Indicator

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #54184

    Sam
    Participant

    This is a solved regression problem from Hunter text book. Here x01 and x02 are different paint suppliers. The intercept constant with standard error given in the book are b01 = 64.9+/-10.2 , b02 = 64.5+/-11.4 and b =11.9+/-4..
    run x01 x02 x y
    1 1 0 2.00 89
    2 0 1 2.50 97
    3 1 0 2.50 91
    4 0 1 2.75 98
    5 0 1 3.00 100
    6 1 0 3.00 104
    7 0 1 3.00 97

    Now the question is when I do regression of this data in Minitab, It will not calculate the intercept constant of x02 also the intercept of x01 is very different. It gives me the error message “””* x02 is highly correlated with other X variables* x02 has been removed from the equation.””
    The Minitab equation is
    y = 64.9 – 0.39 x02 + 11.9 x, Can anyone tell me why the difference between Minitab output and text book results, what am i doing wrong here?

    0
    #194073

    Sam
    Participant

    Dont know how to edit the post… Correction. Minitab equation is y = 64.5 + 0.39 x01 + 11.9 x

    0
    #194075

    Robert Butler
    Participant

    I don’t happen to have BHsquared nearby so I can’t look up the exact problem but what they are doing is generating separate equations for each supplier. Thus the models are y = fn(x01,x) and y = fn(x02,x). What you are trying to do is build a single model y = fn(x01, x02, x) and what Minitab (or any other package for that matter) is telling you is that you can’t because X01 and X02 are confounded.

    0
    #194076

    Sam
    Participant

    Thank you Robert. Now I checked the same problem after reading your explanation and it is clear now.

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

You must be logged in to reply to this topic.