* notes-01feb05.txt

* mreg motivation - case 1:  different Xs (life expectancy)

* mreg motivation - case 2:  prediction with polynomials

TOTAL = b0 + b1*conc + b2*conc^2 + e

TOTAL = b0 + b1*(conc-160) + b2*(conc-160)^2 + e

Can you interpret b1?
*  tough ... can't hold conc^2 constant and vary conc!

What about tests of betas?
*  good stuff!  H0: beta2=0 asks whether you 
need Curvature in a model - given you have a
linear term, do you have more wiggle in the
model

Recall that the t-test of H0: beta2=0

compares ...
TOTAL = b0 + b1*conc + b2*conc^2 + e

TOTAL = b0 + b1*conc + e