Displaying 1 result from an estimated 1 matches for "a2p3".
Did you mean:
a23
2002 Oct 25
0
Age-period-cohort model
...y a newbie question but ...
I try to model age-period-cohort models with polynoms in each effect. I have
for each level of age and period some cases -k- and persons-years -py-.
Models are A_xP_y:
log(k_age,period/py_age,period)=f(polynom(x),age)+f(polynom(y),period). For
exemple I try to adjust an A2P3 :
log(k/py)=a1.age+a2.age*age+p1.period+p2.period*period+p3.period*period*peri
od. I define factor(effect). My command is:
glm(k~-1+a+a2+p+p2+p3+offset(log(py)), family=poisson).
But I obtain only estimation for a and p and "NA" for a2, p2, p3.
What is the problem?
Thanks for advance....