Contains functions for multiple imputation which
complements existing functionality in R.
In particular, several imputation methods for the
mice package (van Buuren & Groothuis-Oudshoorn, 2011,

If you use `miceadds`

and have suggestions for improvement or have found bugs, please email me at [email protected].

The official version of `miceadds`

is hosted on CRAN and may be found here.
The CRAN version can be installed from within R using:

`utils::install.packages("miceadds")`

The version hosted here is the development version of `miceadds`

.
The GitHub version can be installed using `devtools`

as:

`devtools::install_github("alexanderrobitzsch/miceadds")`

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

----------------------- miceadds NEWS -------------------------

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

miceadds: Some additional multiple imputation functions, especially for mice

```
(\-.
/ _`> .---------.
_) / _)= |'-------'|
( / _/ |O O o|
`-.__(___)_ | o O . o |
`---------'
oo__
<;___)------
oo__ " "
<;___)------ oo__
" " <;___)------
" "
```

Alexander Robitzsch, Simon Grund & Thorsten Henke

Questions or suggestions about miceadds should be sent to

Alexander Robitzsch [email protected]

In case of reporting a bug, please always provide a reproducible script.

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

```
CHANGELOG miceadds
```

NOTE * changed default 'inits_lme4=TRUE' to 'inits_lme4=FALSE' (do not use use lme4::lmer() for starting values) in ml_mcmc() NOTE * use optimizer 'bobyqa' by default in multilevel imputations which rely on lme4 package (mice.impute.2l.foo() for foo="pmm","binary" and "continuous" and mice.impute.ml.lmer()) if a 'control' argument is not provided to lme4::lmer() or lme4::glmer() FIXED * fixed a bug in handling arguments in mice.impute.ml.lmer() FIXED * fixed a bug in ma.wtd.statsNA() functions for objects of class BIFIEdata if a statistic for a single variable is requested (thanks to Peter Lenz) NOTE * included argument 'ridge' in ml_mcmc() and ml_mcmc_fit() for stabilizing MCMC sampling in case of convergence issues NOTE * change in computing matches for weighted predictive mean matching (function mice.impute.weighted.pmm()) FIXED * fixed a bug in NMIcombine() with inference of a single parameter NOTE * included example of using mice.impute.pls() at the level of a variable FIXED * fixed a bug in formula handling of ml_mcmc() ADDED * included option for Bayesian bootstrap in mice.impute.pls() by requesting 'use_boot=TRUE' NOTE * included argument 'wt' in pca.covridge() NOTE * exported utility function mice_imputation_get_states() NOTE * included argument 'wy' in mice.impute.weighted.norm(), mice.impute.weighted.pmm(), mice.impute.bygroup() and mice.impute.2lonly.function() NOTE * changed internal handling of grouping variable in mice.impute.bygroup() (due to an example provided by Sandro Tsang; issue #15) ADDED * included substantive model compatible imputation in mice.impute.smcfcs() method NOTE * removed mice.impute.2l.eap() function

DATA * included/modified datasets: --- EXAMP * included/modified examples: mice.impute.ml.lmer (1.2), mice.impute.pls (1), mice.impute.smcfcs (1)

NOTE * included argument '...' in grep.vec() for passing arguments to base::grep() (thanks to Boris Girnat) NOTE * handle now strings of length one in string_to_matrix() FIXED * fixed labeling issue in ma.wtd.corNA() functions when applied to BIFIEdata objects (thanks to Peter Lenz) NOTE * fixed a bug in Example 2 of ?NMIWaldtest caused by changes in mice::pool() output (thanks to Joost van Ginkel) NOTE * included argument 'path2_name' in files_move() ADDED * included arguments 'psi_nu0_list' and 'psi_S0_list' in ml_mcmc() for specification of inverse Wishart prior distribution for covariance matrix of random effects FIXED * fixed a bug in ml_mcmc() when formula arguments was used NOTE * changed default to noninformative prior distributions in ml_mcmc(), i.e. nu0=-3 scale degrees of freedom and zero diagonal scale matrix NOTE * changed default argument values of 'type' in load.data() and load.files() to 'NULL' NOTE * fixed warning message of incorrect argument use of 'imputationMethod' in datalist2mids() (thanks to Valentin Fuchs) NOTE * Replaced all remaining 'imputationMethod' arguments in mice::mice() by 'method' in the manual NOTE * allowed factor variables for PMM imputation methods 'pmm3', 'pmm4', 'pmm5', 'pmm6', '2l.contextual.pmm', '2l.pmm', 'bygroup', '2lonly.function', 'ml.lmer', 'pls', 'tricube.pmm', 'tricube.pmm2' NOTE * imputation method 'tricube.pmm2' is renamed as 'tricube.pmm'. The imputation method 'tricube.pmm2' is now defunct. ADDED * added import format 'xlsx' (or 'xls') in load.data() function using the readxl::read_excel function (package readxl is now included in list of 'Suggests' packages) ADDED * added user-defined data input function in load.data() by including the argument 'load_fun'

DATA * included/modified datasets: --- EXAMP * included/modified examples: filename_split (4), NMIWaldtest (2), mi.anova (1), load.data (1)

NOTE * included separator "\t" in scan.vec() (thanks Thomas Kiefer) FIXED * fixed bug in 'summary' method of glm.cluster() (thanks to tra6sdc; issue #14) NOTE * included utility function load.files() for reading multiple files in a data frame NOTE * corrected a bug in mice.impute.2l.pls2() and in Example 1 of ?mice.impute.pls (thanks to a discussion with Peter Lenz)

DATA * included/modified datasets: --- EXAMP * included/modified examples: scan.vec (1)

NOTE * internal changes in mice.imputation.ml.lmer() caused by application within mice.impute.bygroup() imputation method (thanks to Sophie Stallasch; issue #12) NOTE * renamed ma.wtd.curtosisNA() into ma.wtd.kurtosisNA() ADDED * included simulation function fleishman_sim() for generating non-normal data based on Fleishman polynomial ADDED * included simulation functions nnig_sim() and nnig_coef() for simulating multivariate linearly related non-normally distributed variables (Foldnes & Olsson, 2016)

DATA * included/modified datasets: --- EXAMP * included/modified examples: ma.wtd.statNA (1), fleishman_sim (1), nnig_sim (1)

NOTE * corrected Example 2 in ?micombine.cor (thanks to jamesrrae; #9) FIXED * fixed a bug in mice.impute.bygroup() with two-level imputation models if factor variables are used (see Example 2 in ?mice.impute.bygroup) FIXED * fixed numerical problems when principal component analysis is used for dimension reduction NOTE * included some examples of using mice::make.predictorMatrix() and mice::make.method() functions in the miceadds manual NOTE * included argument 'pls.facs' for PLS dimension reduction in mice.impute.2l.contextual.pmm() and mice.impute.2l.contextual.norm() (communicated by Laura Helbling)

DATA * included/modified datasets: --- EXAMP * included/modified examples: micombine.cor (2), mice.impute.bygroup (2), mice.impute.2l.contextual.pmm (1)

ADDED * included group mean function gm() and centering around group mean function cwc() NOTE * included utility functions ma_lme4_formula_terms() and ma_lme4_formula_design_matrices() for handling lme4 formulas and generating design matrices for mixed effects models FIXED * fixed a bug in mice.impute.weighted.norm() and ...pmm() due to recent changes in mice 3.0 ADDED * added MCMC function ml_mcmc() and ml_mcmc_fit() for estimating mixed effects models. Exported several sub-functions of ml_mcmc(). NOTE * included utility function string_to_matrix() for converting string vectors into matrices NOTE * added utility functions filename_split_vec() and Rcppfunction_remove_classes() ADDED * added functions lmer_vcov() and lmer_pool() for inference of lme4 models for multiply imputed datasets

DATA * included/modified datasets: data.ma08 EXAMP * included/modified examples: GroupMean (2), ma_lme4_formula_terms (1,2), ml_mcmc (1), filename_split(3), Rfunction_include_argument_values (1), lmer_vcov (1,2)

NOTE * internal restructure of functions NMIwaldtest() and MIwaldtest() FIXED * fixed broken functions due to changes in recent CRAN version mice 3.0.0 (communicated by Stef van Buuren): mice.impute.plausible.values(), mice.1chain(), mice.impute.ml.lmer(), mice.impute.2l.pls2() ADDED * added utility function mice_inits(), ma_exists_get(), ma_exists() NOTE * included argument 'remove_empty' in string_extract_part() NOTE * included argument 'cluster_var' for specifying cluster-level variable in mice.impute.2lonly.function()

DATA * included/modified datasets: --- EXAMP * included/modified examples: mice.impute.plausible.values (1), mice_inits (1)

ADDED * included utility Rcpp sourcing function source.Rcpp.all() NOTE * included utility functions Revalpr_maxabs() and Revalpr_round() FIXED * fixed bug in Rfunction_include_argument_values() when some arguments contain parentheses FIXED * fixed a bug in mice.1chain() with implausible input arguments (communicated by Linying Ji) FIXED * corrected a bug in ANSI_create_table() if only one empty row or column should be included NOTE * included utility functions grep_leading(), grepvec_leading() and grepvec() NOTE * included utility function rcpp_create_header_file() for writing cpp header files FIXED * fixed a bug in mice.impute.pmm6() which caused installation problems (thanks to Stef van Buuren) NOTE * changed output in datalist2mids() due to changes of data type of 'visitSequence' in mice (>= 3.0) (requested by Stef van Buuren; Github miceadds ...\6#issuecomment-390218252) NOTE * handle also character input of visit sequence in visitSequence.determine() for coming changes in mice (>= 3.0)

DATA * included/modified datasets: --- EXAMP * included/modified examples: grep.vec (1)

FIXED * fixed a bug in mice.impute.bygroup() if only one predictor variable is used FIXED * fixed a bug in mice.impute.bygroup() whoch was caused by using the recently included "wy" argument in the mice package NOTE * modified mice.impute.bygroup() function for groupwise imputation with multilevel imputation methods ("2l" functions) NOTE * fixed numerical instabilities in plausible values imputation method NOTE * included argument 'control_latreg' in imputation method 'plausible.values' for providing control arguments for function TAM::tam.latreg() NOTE * excluded some packages from list of package 'Suggests'

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

FIXED * replaced the modified pmm6() imputation method causing a bug by a previous one (thanks to Peter Lenz) NOTE * included ridge operator for covariance matrix of predictors in mice.impute.plausible.values() FIXED * fixed a bug in case of perfectly correlated regression coefficients in mice.impute.plausible.values()

DATA * included/modified datasets: --- EXAMP * included/modified examples: mice.impute.plausible.values (3)

NOTE * changed default value of argument 'progress' in datalist2mids() to FALSE NOTE * included utility functions in_CI() and string_extract_part() NOTE * changed names of imputation methods mice.impute.2l.latentgroupmean.ML() and mice.impute.2l.latentgroupmean.MCMC() to mice.impute.2l.latentgroupmean.ml() and mice.impute.2l.latentgroupmean.mcmc() NOTE * changed function name of internal function mice.impute.2l.lmer() to mice_impute_2l_lmer() to avoid confusion because a function mice::mice.impute.2l.lmer() has been recently included in the mice package NOTE * added example for computing global F-test in lm.cluster() (suggested by Thiemo Knaust)

DATA * included/modified datasets: --- EXAMP * included/modified examples: in_CI (1), filenames_split (2), lm.cluster (2)

ADDED * added function jomo2mids() which converts imputation output of the jomo package into an object of class mids FIXED * fixed a bug in mice.impute.2l.pls2() which did not find a C routine in .Call() ADDED * added imputation function mice.impute.ml.lmer() for imputation models based on the linear mixed effects specification in lme4. Hierarchical and non-hierarchical levels are allowed and imputation at higher levels is accomodated. NOTE * added utility functions ma_rmvnorm(), Rfunction_include_argument_values() and Rfunction_output_list_result_function ()

DATA * included/modified datasets: data.ma07 EXAMP * included/modified examples: jomo2datlist (1), mice.impute.ml.lmer (1), ma_rmvnorm (1), Rfunction_include_argument_values (1)

NOTE * removed MASS package from 'Imports'. Moved blme, car, foreign, grouped, MCMCglmm, multiwayvcov and sjlabelled packages from 'Imports' to 'Suggests'. Included hmi and micemd packages in 'Suggests'. NOTE * added comments about hmi and micemd packages in the manual of mice.impute.2l.lmer.Rd. ADDED * added functionality for string variables in 'write.fwf2' (suggested by Tung Nguyen)

DATA * included/modified datasets: --- EXAMP * included/modified examples: write.fwf2 (2)

NOTE * fixed an incorrect description in the manual of 'micombine.F' and 'micombine.chisquare' (thanks to Tobias Rolfes) NOTE * dependency for writing SPSS data to the sjmisc packahe in save.data function has been changed to the sjlabelled package because imported functions were moved from sjmisc to sjlabelled (requested by the maintainer Daniel Luedecke)

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

NOTE * included argument 'draw.fixed' in 'mice.impute.2l.latentgroupmean.MCMC' which allows to specify whether fixed effects parameters should be sampled or not in imputation NOTE * included 'mice.impute.2l.pls2' from previous miceadds versions to ensure backward compatability

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

NOTE * updated incorrect examples in miceadds::micombine.chisquare and miceadds::micombine.F (thanks to Nicholas Breitborde) NOTE * restructured Rcpp functions

DATA * included/modified datasets: --- EXAMP * included/modified examples: micombine.F (1), micombine.chisquare (2)

FIXED * fixed bug in data processing 'mice.impute.pls' and 'mice.impute.plausible.values'. Additional arguments 'envir_pos' and 'extract_data' have been included (thanks to Peter Lenz). FIXED * fixed problems in 'mice.impute.pls' due to derived predictor variables such as interactions due to standard deviations of zero (thanks to Peter Lenz) NOTE * included some utility descriptiptive statistics functions 'stats0', 'mean0', 'sd0', ... NOTE * included correct attributes in 'scale_datlist' function for objects of class datlist (thanks to Andrew Frank-Wilson) NOTE * extended 'scale_datlist' function to handle multiple variables to be standardized

DATA * included/modified datasets: --- EXAMP * included/modified examples: stats0 (1), scale_datlist (3)

NOTE * added utility function 'filename_split' for splitting file names NOTE * added utility function 'files_move' for moving files from one directory into another directory NOTE * included starting values in imputations in the function 'mice.impute.2l.latentgroupmean.MCMC' which are obtained from previous iterations in MCMC imputation chain

DATA * included/modified datasets: --- EXAMP * included/modified examples: filename_split (1), files_move (1)

FIXED * fixed an incorrect calculation in D2 statistic which appeared differently in literature (thanks to David Tian) NOTE * allow now vectors of strings to specify file names in 'load.data'. Also included the value NULL of the argument 'path' if a path should not be specified in this argument. ADDED * included mice imputation method 'mice.impute.hotDeck' for probabilistic hot deck imputation FIXED * fixed a bug included in 'mice.impute.plausible.values' using likelihood objects which was recently introduced in miceadds 1.9 FIXED * fixed a bug in predicted mean matching of two-level data in function 'mice.impute.2l.pmm'

DATA * included/modified datasets: --- EXAMP * included/modified examples: mice.impute.hotDeck (1)

FIXED * fixed a bug in 'mice.impute.2l.binary'. Random slopes were not properly handled in the previuos package version (miceadds 1.8). ADDED * included lme4-based and blme-based imputation of normally distributed two-level data in 'mice.impute.2l.continuous' function. Predictive mean matching for two-level data is also added as the 'mice.impute.2l.pmm' function. ADDED * included imputation function '2lonly.function' for level 2 imputation based on (arbitrary) level 1 imputation function ADDED * added general group-wise imputation function 'mice.impute.bygroup' NOTE * included t values and p values in output of 'pool_mi' for inference of multiply imputed datasets (due to a request on R help list of Jennifer Lorenz) NOTE * renamed PLS imputation method '2l.pls' into 'mice.impute.pls' NOTE * renamed plausible value imputation method into 'mice.impute.plausible.values' NOTE * renamed 'mice.impute.2l.eap' into 'mice.impute.eap' NOTE * Defunct functions: 'mice.impute.2l.pls', 'mice.impute.2l.pls2', 'fast.groupmean', 'fast.groupsum', 'mice.impute.2l.plausible.values', 'mice.impute.2l.eap', 'mice.impute.2lonly.norm2', 'mice.impute.2lonly.pmm2'

DATA * included/modified datasets: --- EXAMP * included/modified examples: mice.impute.2l.lmer (1), mice.impute.2lonly.function (1), mice.imputation.bygroup (1), lm.cluster(1), mice.impute.pls (1)

NOTE * added argument 'subdir' in 'write.datlist' indicating whether output should be written in a subdirectory or directly into the working directory NOTE * fixed a minor bug in creation of Mplus input files in 'write.mice.imputation' and 'write.datlist' NOTE * deleted display of chi square approximation p value in 'micombine.F' and 'micombine.chisquare' (after a discussion with Joanna Kaye) NOTE * extended 'subset' methods for integer input of argument 'subset' which allows easier selection of rows by specifying row indices NOTE * changed dimnames handling in 'NMIwaldtest' which should have no practical side effects ADDED * added utility functions 'ma.wtd.skewnessNA' and 'ma.wtd.curtosisNA' for calculating skewness and curtosis (due to a correspondence with Timo Ehmke) ADDED * added the function 'ma.wtd.quantileNA' for weighted quantile calculation NOTE * extended 'ma.wtd.statNA' functions to handle objects of class 'BIFIEdata' from the BIFIEsurvey package NOTE * renamed 'fast.groupmean' and 'fast.groupsum' into 'GroupMean' and 'GroupSum' and added function 'GroupSD' for calculation of group-wise standard deviations NOTE * replace 'Hmisc::wtd.var' import by 'TAM::weighted_sd' NOTE * shift Hmisc, inline, MBESS and pls packages from Imports to Suggests and removed bayesm from Imports NOTE * rewrite of 'scan.vec' which avoids temporary file savings NOTE * extended 'micombine.cor' to nested multiply imputed datasets, for objects of classes 'datlist' and 'nested.datlist' and single datasets (optionally non-imputed), i.e. the input of 'micombine.cor' must not be mids objects NOTE * added function 'micombine.cov' which has same functionality as 'micombine.cor' NOTE * added attributes in 'datlist' and 'nested.datlist' for cases of varying sample sizes in imputed datasets FIXED * added functionality of handling single imputed datasets in 'pool_mi' function ADDED * added inference for partial correlation in 'micombine.cor' using the argument 'partial' FIXED * fixed an unnecessary warning message in 'datalist2mids' for class checking (thanks to Janet Kwok) ADDED * included imputation methode '2l.binary' of binary variable with two-level logistic regression models

DATA * included/modified datasets: --- EXAMP * included/modified examples: subset.datlist (1), ma.wtd.statsNA (1,2,3,4), GroupMean (1), mi.combine.cor (1,2)

NOTE * allowed alternative specification of weighting variable in
'scale_datlist' which now allows for group-wise
standardizations
NOTE * included arguments 'index' and 'systime' in 'save.data' for
automatic inclusion of indices and time stamps in saved
datasets
NOTE * included argument 'se' in 'NMIcombine' function which can be
used if only standard errors instead of covariance matrices
are available
ADDED * added utility function 'pool_mi' for inference of multiply
imputed datasets which has the same functionality as
'mice::pool' or 'mitools::MIcombine'. This function provides
summary, coef and vcov S3 methods.
NOTE * function 'datalist2mids' is now also accessible via
'datlist2mids' (which is more coherent to class 'datlist')
NOTE * defined 'pool_nmi' synonymously to 'NMIcombine'
NOTE * added S3 print method for 'datlist', 'nested.datlist',
'NestedImputationList'
ADDED * added subsetting methods 'subset_datlist' (and the S3 method
subset) for multiply imputed datasets and 'subset_nested.datlist'
for nested multiply imputed datasets
NOTE * added further calculation of fraction of missing information for
nested multiply imputed datasets according to Harel and
Schafer (2002) in 'NMIcombine' which can be assessed by the
argument 'method=2'.

NOTE * extended imputation of latent group means to lme4-based imputation
in 'mice.impute.2l.latentgroupmean.ML' and MCMCglmm based
imputation in 'mice.impute.2l.latentgroupmean.MCMC'
NOTE * added with and within methods for objects of class 'datlist'
and 'nested.datlist'
ADDED * added convenience functions 'withPool_MI' and 'withPool_NMI' for
pooling (averaging) estimates for (nested) multiply imputed
datasets
ADDED * added utility function 'scan0' which is 'scan' with default
what='character'
ADDED * added utility functions 'nestedList2List' and 'List2nestedList'
for list conversions
ADDED * added functions 'nested.datlist2datlist' and 'datlist2nested.datlist'
for converting nested multiply and multiply imputed datasets
NOTE * extended argument 'type' in 'save.data' for saving a dataset in
multiple file formats
NOTE * added example for plausible value imputation in Amelia package
in draw.pv.ctt.Rd (Example 1)
NOTE * added 'print' methods for 'mids.1chain' and 'mids.nmi' objects
(the method is equivalent to 'summary')
NOTE * exported 'write.fwf2' and 'read.fwf2' for writing and reading
files in fixed width format
NOTE * added utility function 'VariableNames2String' for handling
variable names which should be converted into strings
ADDED * added utility function 'write.datlist' for writing multiply
imputed datasets
ADDED * included function 'mi_dstat' for relating missingness
indicators to covariates

DATA * included/modified datasets: --- EXAMP * included/modified examples: scale_datlist (1), save.data (1), pool.mids.nmi (1,2), pool_mi (1), subset_datlist (1,2), datlist2mids (1), with.miceadds (1,2), scan.vec (1), nestedList2List (1), datlist_create (4), save.data (1), draw.pv.ctt (1), write.fwf2 (1), data.graham (1,2), VariableNames2String (1), write.datlist (1), mi_dstat (1)

NOTE * included argument 'X' in 'mids2datlist' and 'mids2mlwin' which allows the inclusion of further variables in list of multiply imputed datasets NOTE * output of function 'mids2datlist' is now of class 'datlist' or 'nested.datlist' ADDED * extended weighted statistics 'ma.wtd.statNA' to (nested) multiply imputed datasets NOTE * included 'datlist_create' and 'nested.datlist_create' utility function NOTE * changed estimation method in imputation of latent group means in 'mice.impute.2l.latentgroupmean' to MCMC estimation using the MCMCglmm package ADDED * added 'scale_datlist' for adding a standardized variable to a list of multiply imputed datasets NOTE * cleaned namespace file and defined imported functions in a more explicit way

DATA * included/modified datasets: --- EXAMP * included/modified examples: mids2datlist (3), ma.wtd.statNA (1,2,3), datlist_create (1,2), scale_datlist (1,2), cxxfunction.copy (1)

NOTE * changed function 'save.data' due to changes in a recent sjmisc package update on CRAN

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

NOTE * included argument 'suffix' in 'save.data' for easier inclusion of parts of file names when saving data FIXED * fixed a bug in imputation method '2l.latentgroupmean' (reported by Simon Grund) ADDED * added Spearman rank correlation for 'micombine.cor' and an example for statistical inference for Kendalls tau (see 'micombine.cor', Example 2) (suggested by Kathleen Gali)

DATA * included/modified datasets: --- EXAMP * included/modified examples: micombine.cor (2)

ADDED * added imputation method 'grouped' for grouped (coarsed) data
ADDED * included 'NMIcombine' function as an extension to
'MIcombine' (mitools package) to nested multiply
imputed datasets
ADDED * included function 'NMIwaldtest' which performs a multi-
parameter Wald test for nested multiply imputed datasets
NOTE * added argument 'extend' in 'fast.groupmean' which is useful
for the inclusion of group means in original datasets
ADDED * added D1 statistic (Wald test) 'MIwaldtest' as a
multiparameter test for multiply imputed datasets
FIXED * fixed a bug in 'lm.cluster' and 'glm.cluster'.
NOTE * The 'cluster' argument in 'lm.cluster' now allows vector input.
NOTE * included datasets from missing data book of Graham (data.graham)

NOTE * included some examples of the jomo package
NOTE * added argument 'NMI' in 'NMIcombine' function which now also
allows input for non-nested multiple imputation
ADDED * added 'NMIextract' function as an extension of 'MIextract'
(mitools package) to nested multiply imputed datasets
NOTE * added utility function 'library_install' which loads packages
and conducts installations if required

DATA * included/modified datasets: data.ma06, data.graham EXAMP * included/modified examples: mice.impute.grouped (1), pool.mids.nmi (1.4,1.5), NMIwaldtest (1,2), fast.groupmean (1), lm.cluster (1,2), jomo2datlist (1), pool.mids.nmi (1), library_install (1)

NOTE * fixed a bug in 'load.data' for loading datasets of type 'table' ADDED * included utility function 'mids2mlwin' for writing MLwiN files (contributed by Thorsten Luka) NOTE * added suggestions to sav file writing functions in haven and sjmisc packages NOTE * removed dependency on PSPP for writing sav files in 'save.data'. Now, the function 'write_spss' from the sjmisc package is used. NOTE * added further values in 'systime' function ADDED * added a utility function 'visitSequence.determine' for determining visit sequence for an imputation model when passive variables are involved ADDED * added function 'mice.nmi' for conducting nested multiple imputation generating objects of class 'mids.nmi' ADDED * extended 'complete' function from mice package for objects of classes 'mids.1chain' (function 'complete.mids.1chain') and 'mids.nmi' (function 'complete.mids.nmi') ADDED * extended 'mids2datlist' for objects of class 'mids.nmi' ADDED * added 'with' S3 method for objects of class 'mids.1chain' and 'mids.nmi' producing objects of class 'mira' (defined in mice) and 'mira.nmi'. For 'mira.nmi', a summary method is defined. ADDED * added 'within' S3 methods for objects of class 'imputationList' (which are defined in the mitools package) ADDED * added statistical inference for nested multiply imputed datasets in 'pool.mids.nmi' function ADDED * added functions 'lm.cluster' and 'glm.cluster' for cluster robust standard errors for 'lm' and 'glm' functions using the multiwaycov package (and the function 'cluster.vcov' therein) ADDED * added function 'NestedImputationList' function for creating objects of class 'NestedImputationList'. S3 Methods 'with' and 'within' are defined for operating on this list of lists of datasets. The S3 method 'MIcombine' (from mitools package) is extended to objects of class 'Nested ImputationResultList'.

DATA * included/modified datasets: --- EXAMP * included/modified examples: write.pspp (1), save.data (1), systime (1), visitSequence.determine (1), datalist2mids (1), mice.nmi (1), complete.miceadds (1,2), mids2datlist (2), with.miceadds (1,2), pool.mids.nmi (1), lm.cluster (1), NestedImputationList (1)

NOTE * extended 'datalist2mids' function to handle a single dataset ADDED * included plausible value imputation via a provided individual likelihood NOTE * included row.names argument and missing value argumant in 'save.data'

DATA * included/modified datasets: --- EXAMP * included/modified examples: mice.impute.2l.plausible.values (2)

NOTE * fixed a minor bug in write.pspp when a variable only contains values of zero NOTE * allow specification of a part of the file name in 'load.data'; the most recent file is loaded

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

ADDED * included Type 3 sum of squares in ANOVA in 'mi.anova'
(thanks to Florian Buchwald)
ADDED * included a function 'load.Rdata2' which prevents from
the definition of the output in the global environment
as in 'load.Rdata'

ADDED * included utility function 'load.data' and 'save.data'
for loading and saving data files from and to several formats

DATA * included/modified datasets: --- EXAMP * included/modified examples: mi.anova (1), load.Rdata, load.data, save.data

ADDED * included several R utility functions

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

FIXED * included bug fixes for 'mice.impute.2lonly.pmm' (mice version 2.21) as 'mice.impute.2lonly.pmm2' in miceadds -> the same for 'mice.impute.2lonly.norm'

DATA * included/modified datasets: --- EXAMP * included/modified examples: ---

FIXED * fixed a bug in tw.mcmc.imputation for rows with only missings ADDED * included function write.pspp for writing SPSS files using PSPP ADDED * included predict method for kernelpls.fit2 function NOTE * export and extended fast.groupmean function to include weights NOTE * export mice.impute.2l.pls and mice.impute.2l.pls2 FIXED * fixed a bug in the creation of interactions in mice.impute.2l.pls (thanks to Takuya Yanagida) NOTE * included calculation of weighted covariance matrix with missing entries in datasets ADDED * included mice.impute.pmm3, mice.impute.pmm4, mice.impute.pmm5 and mice.impute.pmm6 NOTE * included scale2 as a faster substitute for the scale function in R ADDED * included mice.1chain as an alternative to the mice function which multiply imputes data from a long chain ADDED * included imputation function 'mice.impute.2l.eap' for imputation from a known (normal) posterior distribution ADDED * included function mids2datlist for conversion of a mids object into a list of datasets NOTE * included datasets: all datasets contained in the package NOTE * included examples: for all functions

Versions 0.10 -- 2013-08-18 Versions 0.8 -- 2012-10-18 Versions 0.7 -- 2012-06-18 Versions 0.6 -- 2011-11-29 Versions 0.5 -- 2011-06-02 Versions 0.4 -- 2008-05-31 Versions 0.3 -- 2008-04-07 Versions 0.2 -- 2008-03-27 Versions 0.1 -- 2007-10-21