| |
 |
|
|
Science Forum Index » Statistics - Math Forum » Polytomous logistic regression
Page 1 of 1
|
| Author |
Message |
| Irae119 |
 Posted: Thu Apr 17, 2008 1:14 pm |
|
|
|
Guest
|
Hi,
I have a polytomous logistic regression with a dependant variable with three outcomes: success, negative failure, positive failure. It was a continuous variable put in three categories because below 87% or over 113% the operation is considered a failure. The problematic independent variable is continuous, mean temperature. In the univariate analysis, the OR is near 1 but in the multivariate analysis, the OR jumps to 159.9. What could explain this ?
Thanks in advance ! |
|
|
| Back to top |
|
| Richard Ulrich |
| Posted: Fri Apr 18, 2008 1:08 am |
|
|
|
Guest
|
On Thu, 17 Apr 2008 19:14:52 EDT, Irae119
<sophie_r_l_dubuc@hotmail.com> wrote:
Quote: Hi,
I have a polytomous logistic regression with a dependant variable with three outcomes: success, negative failure, positive failure. It was a continuous variable put in three categories because below 87% or over 113% the operation is considered a failure. The problematic independent variable is continuous, mean temperature. In the univariate analysis, the OR is near 1 but in the multivariate analysis, the OR jumps to 159.9. What could explain this ?
Thanks in advance !
Are you saying that the linear trend is not-significant,
whereas the quadratic possibility (3 groups) has a
high OR? That should not be hard to demonstrate
from a graph.
--
Rich Ulrich
http://www.pitt.edu/~wpilib/index.html |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Thu Jul 24, 2008 5:41 pm
|
|
