Logistic regression. biological significance of a mathematical model quadratic term.

#1
See if anybody can help me understand something, as I am a bit confused. It is regarding the biological significance of a mathematical model quadratic term. I am using an Ordinary Logistic Regression Model to model the probability of growth depending on 3 parameters (x, y and z).
I am using the follwing equation:
1/1-P=a0+a1.x+a2.y+a3.z+a4.x.y+a5.x.z+…
The results indicate that significative parameters are a2, a3, a4 and a7.
Now, a7 is a quadratic term; that parameter is food intake.
My question is: How do I interpret biologically that food intake^2 is significant but not significant when it's not quadratic?How do I explain that?
Thanks very much.
 

Dason

Ambassador to the humans
#2
Re: Logistic regression. biological significance of a mathematical model quadratic te

Please only post a single thread. Thank you.

Are you saying that the quadratic term for a variable is significant in the full model but the linear term isn't? Is this what is causing you concern?
 
#3
Re: Logistic regression. biological significance of a mathematical model quadratic te

Sorry about the double post, i posted it in the wrong place the first time so had to re-post it in the correct section.

Yes, that is exactly my concern. Why the term is significant when it is quadratic and it isnt when linear. How would I interpret this in a 'biological' sense?

Also...I have another query...SAS analysis yields this term as significant but I don't understand if the significance relates to the estimate or to the effect of the parameter over growth...

Thank you.