NEWS.md file to track changes to the package.cooks.distance() and mdffits() with include.attr = TRUE now returns a tibble with column names for the fixed effects in agreement with the fixef() output.hlm_resid() output is now .id to avoid conflicts with id columnsAdded separate functions to calculate residuals for LMEs.
resid_marginal() calculates (raw, studentized, Pearson, or Cholesky) marginal residuals.resid_conditional calculates (raw, studentized, Pearson, or Cholesky) conditional residuals (i.e., error terms).resid_ranef calculates raw and standardized predicted random effects.The following functions are now defunct
HLMresiddiagnosticsgroup_qqnormggplot_qqnormcase_delete function and accompaning functions for lme objects from the nlme package. These diagnostics are also avaliable for three-level models.hlm_resid for lme model objects and three-level models.group parameter for influence diagnostics has been changed to level in order to match the residual functions.
level defaults to NULL, which will delete individual observations iteratively as group = NULL did.cooks.distance and mdffits functions now only return the values as a numeric vector, instead of also returning the
beta values as attributes. If these attributes are desired, the user can now set include.attr = TRUE, and a tibble
will be returned instead with the influence diagnostics and the beta attributes.dotplot_diag function has been updated to be more efficient. Additionally, it no longer places labels on the y-axis
and only labels the top five observations in order to improve visibility.LSresid now returns only the residual values, excluding the model framesim argument for LSresid has been removedhlm_influence function has been added. This function returns influence diagnostics appended to the model frame.hlm_resid function has been added. This function returns residual diagnostics appended to the model frame.hlm_augment function has been added, which combines hlm_influence and hlm_resid to return influence diagnostics and residuals appended to the model frame.pull_resid function has been added. This funciton returns a vector of a specified type of residual prioritizing computational efficiency.delete parameter in case_delete, hlm_influence, and hlm_augment now also accepts character vectors at the second or third level. Observations or groups to be deleted can be specified by row indices in a numeric vector (as previously), or as character vectors of group level names found in model@flist (lmerMod models) or model$groups (lme models).cooks.distance and mdffits functions were fixed to solve an issue with the number of columns.case_delete so that it works with three level models.hlm_resid, hlm_influence, and hlm_augment properly respect na.action and work with models fit in nlmeLSresids doesn't break with three-level models, or with models containing tranformed variables.extractV.lme (and thus .lme_matrices) was fixed to work with more complex covariance structures fit via nlme.ggplot2lmer function in lme4or thelmefunction innlme. I am still working to implement these methods for higher-level models using lme`.rotate_ranef function has been added. This function rotates the random
effects in an effort to find the least confounded residuals for distributional
assessment.LSresids was fixed for an issue with the order of the resulting data frame.case_delete was fixed so that numeric group labels work properly, which fixes an issue
with rvc.group_qqnorm, by using ppoints rather than .SampleQuantiles.case_delete.lmerMod, to use the getME() function to extract n.HLMresid
to a more appropriate definition: e / var(\hat{e}).group_qqnormcooks.distance, mdffits, covratio,
covtrace, rvc, and leverage for objects of class mer.case_delete function, with
corresponding S3 methods for objects of class case_delete.leverage function has been added.case_delete and other deletion functions now allow for the user to
manually specify a subset to delete.dotplot_diag to accomodate a more general
usage.diagnostics no longer requires a model parameter to be specified.delete parameter has been added to case_delete to allow for manual
specification of a subset that should be deleted.case_delete the user should specify
group = NULL rather than group = FALSE.