Skip to content
Prev 17017 / 20628 Next

Significance of B-splines components in mixed-effects logistic regression (glmer)

Dear Anne,

For comparison, here's a GAMM for the same example:

# Anne's code first, then

library(gamm4)
gam.mod <- gamm4(DV ~ s(IV1) + IV2 + IV3, random=~(1|RV1) + (1|RV2),
???????????????? data=df, family=binomial)
summary(gam.mod$mer,corr=FALSE)
summary(gam.mod$gam)

With the exception of the intercepts and the smooth/spline parameters
(which are by definition different), the estimates are quite similar.
The smoother is also estimated to be roughly linear (edf=1), so it's not
surprising that it and also your splines aren't significant.

I'm not sure if languageR::dative was just a convenient example, but if
you are working in psycholinguistics or cognitive neuroscience, GA(M)Ms
are starting to gain traction:

Baayen, H., Vasishth, S., Kliegl, R., & Bates, D. (2017). The cave of
shadows: Addressing the human factor with generalized additive mixed
models. Journal of Memory and Language , 94 , 206?234.
doi:10.1016/j.jml.2016.11.006

Tremblay, A., & Newman, A. J. (2015). Modeling nonlinear relationships
in ERP data using mixed-effects regression with R examples.
Psychophysiology , 52 , 124?139. doi:10.1111/psyp.12299

Wieling, M., Nerbonne, J., & Baayen, R. H. (2011). Quantitative social
dialectology: explaining linguistic variation
geographically and socially. PLoS One, 6(9), e23613.
doi:10.1371/journal.pone.0023613

Phillip
On 21/09/18 19:04, Anne Lerche wrote: