These methods let emmeans build reference grids for mm_lmm and
mm_glmm objects when the optional emmeans package is installed. They
expose the same fixed-effect design surface used by mm_grid() and
mm_means().
Usage
recover_data.mm_lmm(object, data = NULL, ...)
emm_basis.mm_lmm(
object,
trms,
xlev,
grid,
method = c("auto", "satterthwaite", "kenward_roger", "asymptotic", "none"),
...
)
recover_data.mm_glmm(object, data = NULL, ...)
emm_basis.mm_glmm(object, trms, xlev, grid, ...)Value
Objects expected by emmeans::recover_data() and
emmeans::emm_basis().
Details
The bridge covers population fixed-effect means only. For GLMMs the whole
emmeans surface is gated by the package's inference-capability contract:
a profiled (pirls_profiled) fit refuses with a typed error (refit with
method = "joint_laplace" for certified Wald inference, or opt in to the
labelled working-Hessian approximation at fit time with
inference = "working_hessian"). When the fitted artifact carries a
usable mixedmodels.fixed_effect_covariance_matrix payload, emmeans
receives that full fixed-effect covariance matrix. Native
mm_predictions(), mm_means(), and mm_comparisons() remain the
contract-preserving mixeff surface for LMMs because they preserve
row-level status and reason fields.