Case Deletion for mer/lmerMod objects — case

This function is used to iteratively delete groups corresponding to the levels of a hierarchical linear model. It uses lmer() to fit the models for each deleted case (i.e. uses brute force). To investigate numerous levels of the model, the function will need to be called multiple times, specifying the group (level) of interest each time.

# Default S3 method
case_delete(model, ...)

# S3 method for class 'mer'
case_delete(
  model,
  level = 1,
  type = c("both", "fixef", "varcomp"),
  delete = NULL,
  ...
)

# S3 method for class 'lmerMod'
case_delete(
  model,
  level = 1,
  type = c("both", "fixef", "varcomp"),
  delete = NULL,
  ...
)

# S3 method for class 'lme'
case_delete(
  model,
  level = 1,
  type = c("both", "fixef", "varcomp"),
  delete = NULL,
  ...
)

Arguments

model: the original hierarchical model fit using lmer()
...: do not use
level: a variable used to define the group for which cases will be deleted. If level = 1 (default), then the function will delete individual observations.
type: the part of the model for which you are obtaining deletion diagnostics: the fixed effects ("fixef"), variance components ("varcomp"), or "both" (default).
delete: numeric index of individual cases to be deleted. If the level parameter is specified, delete may also take the form of a character vector consisting of group names as they appear in flist. It is possible to set level and delete individual cases from different groups using delete, so numeric indices should be double checked to confirm that they encompass entire groups. If delete = NULL then all cases are iteratively deleted.

Value

a list with the following components:

fixef.original: the original fixed effects estimates
ranef.original: the original predicted random effects
vcov.original: the original variance-covariance matrix for the fixed effects
varcomp.original: the original estimated variance components
fixef.delete: a list of the fixed effects estimated after case deletion
ranef.delete: a list of the random effects predicted after case deletion
vcov.delete: a list of the variance-covariance matrices for the fixed effects obtained after case deletion
fitted.delete: a list of the fitted values obtained after case deletion
varcomp.delete: a list of the estimated variance components obtained after case deletion

References

Christensen, R., Pearson, L.M., and Johnson, W. (1992) Case-Deletion Diagnostics for Mixed Models, Technometrics, 34, 38 – 45.

Schabenberger, O. (2004) Mixed Model Influence Diagnostics, in Proceedings of the Twenty-Ninth SAS Users Group International Conference, SAS Users Group International.

Author

Adam Loy loyad01@gmail.com

Examples

data(sleepstudy, package = 'lme4')
fm <- lme4::lmer(Reaction ~ Days + (Days|Subject), sleepstudy)

# Deleting every Subject
fmDel <- case_delete(model = fm, level = "Subject", type = "both")

# Deleting only subject 308
del308 <- case_delete(model = fm, level = "Subject", type = "both", delete = 308)

# Deleting a subset of subjects
delSubset <- case_delete(model = fm, level = "Subject", type = "both", delete = 308:310)