Predictions of a linear mixed model fitted by lmer, at the rows of the fit or at new data, with all, some
or none of the random-effects terms, optionally with their standard errors. Equivalent to lme4::predict() in R.
The prediction of a row is x'β̂ + Σ z'b̂ over the selected terms, x built from the row as the fit built its
fixed-effects matrix (its terms, contrasts, factor levels and dropped columns) and b̂ the conditional mode of the
row's level. A missing value in a variable gives NaN for its row; a factor level of the fixed part that the fit did
not have is the RangeError factor <name> has new level <value> (R: non-conformable arguments). The standard
errors are those of the conditional covariance σ² A K⁻¹ A' of (b, β) at θ̂ (Henderson's equations), computed
correctly also where lme4's are wrong (crossed grouping factors, some reForm subsets).
Predictions of a linear mixed model fitted by lmer, at the rows of the fit or at new data, with all, some or none of the random-effects terms, optionally with their standard errors. Equivalent to
lme4::predict()in R.The prediction of a row is x'β̂ + Σ z'b̂ over the selected terms, x built from the row as the fit built its fixed-effects matrix (its terms, contrasts, factor levels and dropped columns) and b̂ the conditional mode of the row's level. A missing value in a variable gives NaN for its row; a factor level of the fixed part that the fit did not have is the RangeError
factor <name> has new level <value>(R:non-conformable arguments). The standard errors are those of the conditional covariance σ² A K⁻¹ A' of (b, β) at θ̂ (Henderson's equations), computed correctly also where lme4's are wrong (crossed grouping factors, somereFormsubsets).