search for: sd_b1

...convey the uncertainty of this function (e.g. computing the confidence interval of the difference at each value of C). One thought is that predict.gam(...,se.fit=T) returns SE values, so if I could find out the N on which these SE values are computed, I could compute the difference CI as sqrt( ( (SD_B1)^2 + (SD_B2)^2 ) / N ) * qt( .975, df=N-1 ) However, I can't seem to figure out what value of N was used to compute the SEs that predict.gam(...,se.fit=T) produces. Can anyone point me to where I might find N? Further, is N-1 the proper df for the call to qt()? Finally, with a smooth functio...