COMPARING PARAMETRIC AND GAM FITS

[Link Map]
begin comments text sound
line

Comparison between the models is given below 
Model                       Residual df(err) Change in Change 
                            Deviance         Deviance   in df

s(age) + s(start)            50.4     74.1      
poly(age,2) + s(start)       50.6     75.0     0.2      0.9
poly(age,2) + poly(start,3)  51.2     75.0     0.8      0.9
poly(age,2) 
  + (start-12)*I(start>12)   51.6     75.0     1.2      0.9
The Chi-squared value for 1 degree of freedom at the 5% level is 3.84, so it can be seen that each of the models listed above is a good approximation to the original non-parametric model.

line
Generalized Additive Models 23.4.96 Page : 12d of 15