Dear all, I have 10 studies. I want to fit a generalized additive model for each of 8 studies. For combining the results from 10 studies, I have to specify the same knots (30,40,50,60,70,80,90) and knots boundary, 20 and 100, in gam(mgcv). Now I just found knots= option in the gam function, so I added the boundary of age=20 and age=100 into knots list. data1.gam<-gam(y~male+s(age,by=as.factor(male),bs='cr',k=9), data=data1, family=binomial,knots=list(age=c(20,30,40,50,60,70,80,90,100))) Since the number of knots is 9, I have to set k=9 for 'cr 'basis spline. data1.gam$coeff gives 8 coefficients for each of two gender groups, male and female. I am wondering how to specify the same knots and their boundary and 'cr' is appropriate for this case. In addition, why is there 8 coefficients given instead of 9? I am a beginner of GAM. Only a week. Could someone please give a help. Thanks a lot. Sincerely, Yan [[alternative HTML version deleted]]