cmp_R2 for crossed random effect models?

[bbolker]

cmp_R2 has the signature cmp_R2(c=C, x=X, SigHat=SigHat, beta=beta, obsperclust = obsperclust, nclusts = nclusts, method = 'sgv'). I'm trying to apply it to a crossed random effects model fitted with gamm4 (I'm extracting the $mer component of the fitted model, so it's a merMod object).

I understand how to get the contrasts matrix; I'm getting SigHatwith the utility function below (extracted from the body of r2beta.lmerMod). I'm a little confused by obsperclust, nclusts, which would only seem to be unambiguously defined for a model with a single grouping variable.

I know I can go look at the papers to try to figure it out (and I will), but was hoping for a hint/some guidance ...

---

extract.merMod.cov <- function(model) {
    Z = lme4::getME(model, "Z")
    s2e = lme4::getME(model, "sigma")^2
    lam = lme4::getME(model, "Lambda")
    lamt = lme4::getME(model, "Lambdat")
    G = s2e * (lam %*% lamt)
    SigHat = Z %*% (G %*% Matrix::t(Z))
    Matrix::diag(SigHat) = Matrix::diag(SigHat) + s2e/model@resp$weights
    return(SigHat)
}