Description

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, <doi:10.18637/jss.v045.i03>) are included. Main features of the miceadds package include plausible value imputation (Mislevy, 1991, <doi:10.1007/BF02294457>), multilevel imputation for variables at any level or with any number of hierarchical and non-hierarchical levels (Grund, Luedtke & Robitzsch, 2018, <doi:10.1177/1094428117703686>; van Buuren, 2018, Ch.7, <doi:10.1201/9780429492259>), imputation using partial least squares (PLS) for high dimensional predictors (Robitzsch, Pham & Yanagida, 2016), nested multiple imputation (Rubin, 2003, <doi:10.1111/1467-9574.00217>), substantive model compatible imputation (Bartlett et al., 2015, <doi:10.1177/0962280214521348>), and features for the generation of synthetic datasets (Reiter, 2005, <doi:10.1111/j.1467-985X.2004.00343.x>; Nowok, Raab, & Dibben, 2016, <doi:10.18637/jss.v074.i11>).

Details

• The miceadds package contains some functionality for imputation of multilevel data. The function mice.impute.ml.lmer is a general function for imputing multilevel data with hierarchical or cross-classified structures for variables at an arbitrary level. This imputation method uses the lme4::lmer function in the lme4 package. The imputation method mice.impute.2lonly.function conducts an imputation for a variable at a higher level for already defined imputation methods in the mice package. Two-level imputation is available in several functions in the mice package (mice::mice.impute.2l.pan, mice::mice.impute.2l.norm) as well in micemd and hmi packages. The miceadds package contains additional imputation methods for two-level datasets: mice.impute.2l.continuous for normally distributed data, mice.impute.2l.pmm for predictive mean matching in multilevel models and mice.impute.2l.binary for binary data.

• In addition to the usual mice imputation function which employs parallel chains, the function mice.1chain does multiple imputation from a single chain.

• Nested multiple imputation can be conducted with mice.nmi. The function NMIcombine conducts statistical inference for nested multiply imputed datasets.

• Imputation based on partial least squares regression is implemented in mice.impute.pls.

• Unidimensional plausible value imputation for latent variables (or variables with measurement error) in the mice sequential imputation framework can be applied by using the method mice.impute.plausible.values.

• Substantive model compatible multiple imputation using fully conditional specification can be conducted with mice.impute.smcfcs.

• The function syn_mice allows the generation of synthetic datasets with imputation methods for mice. It has similar functionality as the synthpop package (Nowok, Raab, & Dibben, 2016). The function mice.impute.synthpop allows the usage of synthpop synthesization methods in mice, while syn.mice allows the usage of mice imputation methods in synthpop.

• The method mice.impute.simputation is a wrapper function to imputation methods in the simputation package. The methods mice.impute.imputeR.lmFun and mice.impute.imputeR.cFun are wrapper functions to imputation methods in the imputeR package.

• The miceadds package also includes some functions R utility functions (e.g. write.pspp, ma.scale2).

• Imputations for questionnaire items can be accomplished by two-way imputation (tw.imputation).

Author(s)

Alexander Robitzsch [aut,cre] (<https://orcid.org/0000-0002-8226-3132>), Simon Grund [aut] (<https://orcid.org/0000-0002-1290-8986>), Thorsten Henke [ctb]

Maintainer: Alexander Robitzsch <robitzsch@ipn.uni-kiel.de>

References

Bartlett, J. W., Seaman, S. R., White, I. R., Carpenter, J. R., & Alzheimer's Disease Neuroimaging Initiative (2015). Multiple imputation of covariates by fully conditional specification: Accommodating the substantive model. Statistical Methods in Medical Research, 24(4), 462-487. doi: 10.1177/0962280214521348

Grund, S., Luedtke, O., & Robitzsch, A. (2018). Multiple imputation of multilevel data in organizational research. Organizational Research Methods, 21(1), 111-149. doi: 10.1177/1094428117703686

Mislevy, R. J. (1991). Randomization-based inference about latent variables from complex samples. Psychometrika, 56(2), 177-196. doi: 10.1007/BF02294457

Nowok, B., Raab, G., & Dibben, C. (2016). synthpop: Bespoke creation of synthetic data in R. Journal of Statistical Software, 74(11), 1-26. doi: 10.18637/jss.v074.i11

Reiter, J. P. (2005) Releasing multiply-imputed, synthetic public use microdata: An illustration and empirical study. Journal of the Royal Statistical Society, Series A, 168(1), 185-205. doi: 10.1111/j.1467-985X.2004.00343.x

Robitzsch, A., Pham, G., & Yanagida, T. (2016). Fehlende Daten und Plausible Values. In S. Breit & C. Schreiner (Hrsg.). Large-Scale Assessment mit R: Methodische Grundlagen der oesterreichischen Bildungsstandardueberpruefung (S. 259-293). Wien: facultas.

Rubin, D. B. (2003). Nested multiple imputation of NMES via partially incompatible MCMC. Statistica Neerlandica, 57(1), 3-18. doi: 10.1111/1467-9574.00217

van Buuren, S. (2018). Flexible imputation of missing data. Boca Raton: CRC Press. doi: 10.1201/9780429492259

van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate imputation by chained equations in R. Journal of Statistical Software, 45(3), 1-67. doi: 10.18637/jss.v045.i03

See Also

See also the CRAN task view Missing Data:
https://CRAN.R-project.org/view=MissingData

See other R packages for conducting multiple imputation: mice, Amelia, pan, mi, norm, norm2, BaBooN, VIM, ...

Some links to internet sites related to missing data:

http://missingdata.lshtm.ac.uk/
http://www.stefvanbuuren.nl/mi/
http://www.bristol.ac.uk/cmm/software/realcom/
https://rmisstastic.netlify.com/

Examples