# Variance and multiple levels in clmm

I'm trying to fit a multi-level model where people are nested in groups. V1 is my outcome variable. G2 is my highest level group and G1 is a group nested within G2

m1 <- clmm(V1 ~ 1 + (1 | G2), data = df, Hess = T)
m2 <- clmm(V1 ~ 1 + (1 | G1) + (1 | G2), data = df, Hess = T)


Based on AIC values, m1 is the best fit, so I'll go from there.

m3 <- clmm(V1 ~ 1 + (1 | G1:G2), data = df, Hess = T)
m4 <- clmm(V1 ~ 1 + (1 | G2/G1), data = df, Hess = T)


Is the syntax for m3 nesting G2 in G1 or G1 in G2? What about for m4?

Lastly, is the variance produced by summary(m1), summary(m2) etc. the variance that remains in the model? If I do a cross level interaction between G1:G2, how should I be interpreting the variance in the RE effects portion of the results?

             Var   Std.Dev
G1:G2    9.017e-11 9.496e-06
G1       1.683e-01 4.103e-01


I'm sorry for the fundamental-ness of this question and the lack of data.