• Added a 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.
  • First columns of hlm_resid() output is now .id to avoid conflicts with id columns

NEW FEATURES

Added 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.

USER-VISIBLE CHANGES

The following functions are now defunct

  • HLMresid
  • diagnostics
  • group_qqnorm
  • ggplot_qqnorm

DEVELOPMENT

  • Influence diagnostics are now avaliable through the case_delete function and accompaning functions for lme objects from the nlme package. These diagnostics are also avaliable for three-level models.
  • Residual diagnostics are also now available via hlm_resid for lme model objects and three-level models.

USER-VISIBLE CHANGES

  • The 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.
  • The 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.
  • The 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 frame
  • The sim argument for LSresid has been removed

NEW FEATURES

  • The hlm_influence function has been added. This function returns influence diagnostics appended to the model frame.
  • The hlm_resid function has been added. This function returns residual diagnostics appended to the model frame.
  • The hlm_augment function has been added, which combines hlm_influence and hlm_resid to return influence diagnostics and residuals appended to the model frame.
  • The pull_resid function has been added. This funciton returns a vector of a specified type of residual prioritizing computational efficiency.
  • The 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).

BUG FIXES

  • cooks.distance and mdffits functions were fixed to solve an issue with the number of columns.
  • Fixed an issue with 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 nlme
  • LSresids doesn’t break with three-level models, or with models containing tranformed variables

BUG FIXES

  • .extractV.lme (and thus .lme_matrices) was fixed to work with more complex covariance structures fit via nlme.
  • Updated package to work with the most recent version of ggplot2

DEVELOPMENT

  • Influence diagnostics in HLMdiag 0.3.0 are available for two-level models fit using the lmer function in lme4or thelmefunction innlme. I am still working to implement these methods for higher-level models usinglme`.
  • HLMdiag no longer loads lme4 automatically (see above for the reason).

NEW FEATURES

  • The rotate_ranef function has been added. This function rotates the random effects in an effort to find the least confounded residuals for distributional assessment.

BUG FIXES

  • 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.
  • Fixed an issue with group_qqnorm, by using ppoints rather than .SampleQuantiles.
  • Fixed an issue with case_delete.lmerMod, to use the getME() function to extract n.
  • A bug in the calculation of the Cholesky residuals was fixed (thanks to Harry Hiemstra for reporting the bug and the fix)
  • Fixed a compatibility issue with Rcpp
  • Added citation for the JSS paper
  • Fixed a bug with the calculation of Cook’s distance
  • Added a function to calculate rotated random effects
  • Added new data sets
  • Maintenance for compatibility with lme4 1.0 and R 3.0.2
  • Changed the standardization of the EB level-1 residuals in HLMresid to a more appropriate definition: e / var().
  • Added functions to add compatibility with the development version of lme4.
  • Fixed a bug in group_qqnorm
  • Checked compatibility with R 2.15.3

DEVELOPMENT

  • Influence diagnostics in HLMdiag 0.2.0 are compatible with hierarchical (multilevel) linear models of any size and with models with crossed factors.
  • HLMdiag 0.2.0 offers significantly faster computation of the deletion diagnostics for fixed effects that are based on one-step approximations.
  • S3 methods have been created for cooks.distance, mdffits, covratio, covtrace, rvc, and leverage for objects of class mer.
  • Full deletions are still available using the case_delete function, with corresponding S3 methods for objects of class case_delete.

NEW FEATURES

  • A leverage function has been added.
  • case_delete and other deletion functions now allow for the user to manually specify a subset to delete.

OTHER USER-VISIBLE CHANGES

  • changes to the arguments of dotplot_diag to accomodate a more general usage.
  • diagnostics no longer requires a model parameter to be specified.
  • A delete parameter has been added to case_delete to allow for manual specification of a subset that should be deleted.
  • For observation-level deletion using case_delete the user should specify group = NULL rather than group = FALSE.
  • Updates for compatibility with ggplot2 >= 0.9.2
  • Updates to NAMESPACE
  • Updates to NAMESPACE to fix compatibility issues
  • Updates for compatibility with ggplot2 0.9.0
  • Added “marginal” residuals to the type argument for HLMresid.

USER-VISIBLE CHANGES:

  • Removed the formula argument from LSresids. The formula is now obtained automatically from the mer object. Note: we are still working on automatic recognition of math operators such as log(), but anything in I() is recognized.
  • Added the function HLMresid, a wrapper that will extract both the LS and EB residuals given an mer object.
  • Added a level argument to LSresids, so the function can extract LS residuals from either level of the model.

BUG FIXES:

  • Fixed the ordering of output from ‘LSresids’ to match the mode frame obtained from the mer object.
  • Created ‘adjust_lmList’ class to handle fitting separate linear models when a factor is constant across the group.
  • ‘random_ls_coef’ was removed and replaced by the ‘coef’ method for ‘adjust_lmList’ objects.
  • Added ‘rvc’ diagnostic.
  • Extended ‘case_delete’ from only handling deletion for fixed effects to also handling deletion for variance components.
  • Improved ‘dotplot_diag’ to handle modified dotplots.