hlm_augment {HLMdiag}

Calculating residuals and influence diagnostics for HLMs hlm_augment is used to compute residuals, fitted values, and influence diagnostics for a hierarchical linear model. The residuals and fitted values are computed using Least Squares(LS) and Empirical Bayes (EB) methods. The influence diagnostics are computed through one step approximations.

Description

Calculating residuals and influence diagnostics for HLMs

hlm_augment is used to compute residuals, fitted values, and influence diagnostics for a hierarchical linear model. The residuals and fitted values are computed using Least Squares(LS) and Empirical Bayes (EB) methods. The influence diagnostics are computed through one step approximations.

Usage

hlm_augment(object, ...)

## Default S3 method:
hlm_augment(object, ...)

## S3 method for class 'lmerMod'
hlm_augment(object, level = 1, include.ls = TRUE, data = NULL, ...)

Arguments

object

an object of class lmerMod or lme.

...

currently not used

level

which residuals should be extracted and what cases should be deleted for influence diagnostics. If level = 1 (default), then within-group (case-level) residuals are returned and influence diagnostics are calculated for individual observations. Otherwise, level should be the name of a grouping factor as defined in flist for a lmerMod object or as in groups for a lme object. This will return between-group residuals and influence diagnostics calculated for each group.

include.ls

a logical indicating if LS residuals should be included in the return tibble. include.ls = FALSE decreases runtime substantially.

data

the original data frame passed to 'lmer'. This is only necessary for 'lmerMod' models where 'na.action = "na.exclude"'

Details

The hlm_augment function combines functionality from hlm_resid and hlm_influence for a simpler way of obtaining residuals and influence diagnostics. Please see ?hlm_resid and ?hlm_influence for additional information about the returned values.

Note

hlm_augment does not allow for the deletion of specific cases, the specification of other types of leverage, or the use of full refits of the model instead of one step approximations for influence diagnostics. If this additional functionality is desired, hlm_influence should be used instead. The additional parameter standardize is available in hlm_resid; if this are desired, hlm_resid should be used instead.