In #6, there was a wrapper added to perform the extraction. However, we extracted a bit too many terms. In particular, we only want to extract elements corresponding to the random-effects, not all effects (p). c.f.

Consider revising given:

Issue arose with:

m <- lmer(DV ~ P + P:H + (1+P||id), data = data)

No data was given, so an equivalent test case should be:

fm1 <- lmer(Reaction ~ Days + (1+Days || Subject), data = sleepstudy)

This data set has been missing from the package since release.

Remove src/init.c

Regenerate with Rcpp::compileAttributes()

The icc_beta threw the following error:

error: Mat::row(): index out of bounds
Error in icc_beta(X, df_icc[, "L2id"], T1, vy) : 
  Mat::row(): index out of bounds

I just tried to calculate the icc_beta for a dataset with the following structure:
I have a dataset where L1 = time point, L2 = person. Each L2 unit has several L1 units.

However,

• the L1 are not equal between L2 (different group sizes).
• and sometimes persons skipped time points, resulting in cases where L1 may be NA for, say, L1id 1, 2, 3, but have a value for L1id 4,5,6. In such cases, the L1id rows 1,2,3 are retained, even if NA for the Y variable.

My hunch is that this data structure leads to the aforementioned error. Specifically, df_icc[, "L2id"] is 444, whereas nrow(X) is 297.

New R check is in R-devel (target 4.0 in April)

Variables with usage in documentation object ‘dataset’ but not in code

Fixed by adding LazyData: true to DESCRIPTION in #7.

Need to deploy to CRAN

Instead of forcing the user to decompose the model, we'll retrieve the model components automatically and, then, compute the icc.