[R-sig-ME] Reporting coefficients from model averaged GAM (M) models with smoothed terms

Gitu wa Mbui gitumbui at gmail.com
Fri Feb 12 11:44:59 CET 2016

Having run several GAM (M) models with smoothed terms in model selection, I
am wondering how best to report the coefficients (standardized) associated
with each predictor - where each predictor is repeated across >1 models. I
have used model averaging (MuMIn package) to average the models but the
coefficients are returned for each knot rather than overall for each
predictor (for example, when k=5, it returns coefficient for k1 to k4). Is
it Ok to report averaged coefficients regardless of the knot (k) level, and
how would you go about obtaining these?


	[[alternative HTML version deleted]]

More information about the R-sig-mixed-models mailing list