Package 'miceadds'

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 implemented. 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 <[email protected]>

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

##
##   ::'''''''''''''''''''''''''''''''''::
##   :: miceadds 0.11-69 (2013-12-01)   ::
##   ::'''''''''''''''''''''''''''''''''::
##
##  ----------------------- mice at work ---------------------------------
##
##                         (\-.
##                         / _`> .---------.
##                 _)     / _)=  |'-------'|
##                (      / _/    |O   O   o|
##                 `-.__(___)_   | o O . o |
##                               `---------'
##
##                                          oo__
##                                         <;___)------
##                                    oo__   " "
##                                   <;___)------     oo__
##                                     " "           <;___)------
##                                                     " "

Description

Creates imputed dataset from a mids.nmi or mids.1chain object.

Usage

## S3 method for class 'mids.nmi'
complete(data, action=c(1,1), ...)

## S3 method for class 'mids.1chain'
complete(data, action=1, ...)

Arguments

See Also

See also the corresponding mice::complete function and mitml::mitmlComplete.

Imputation methods: mice.nmi, mice.1chain

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Nested multiple imputation and dataset extraction for TIMSS data
#############################################################################

library(BIFIEsurvey)
data(data.timss2, package="BIFIEsurvey" )
datlist <- data.timss2

# remove first four variables
M <- length(datlist)
for (ll in 1:M){
    datlist[[ll]] <- datlist[[ll]][, -c(1:4) ]
}

#***************
# (1) nested multiple imputation using mice
imp1 <- miceadds::mice.nmi( datlist,  m=4, maxit=3 )
summary(imp1)

#***************
# (2) nested multiple imputation using mice.1chain
imp2 <- miceadds::mice.nmi( datlist, Nimp=4, burnin=10,iter=22, type="mice.1chain")
summary(imp2)

#**************
# extract dataset for third orginal dataset the second within imputation
dat32a <- miceadds::complete.mids.nmi( imp1, action=c(3,2) )
dat32b <- miceadds::complete.mids.nmi( imp2, action=c(3,2) )

#############################################################################
# EXAMPLE 2: Imputation from one chain and extracting dataset for nhanes data
#############################################################################

library(mice)
data(nhanes, package="mice")

# nhanes data in one chain
imp1 <- miceadds::mice.1chain( nhanes, burnin=5, iter=40, Nimp=4,
            method=rep("norm", 4 ) )

# extract first imputed dataset
dati1 <- miceadds::complete.mids.1chain( imp1, action=1 )

## End(Not run)

Description

This function removes CF line endings from a text file and writes the processed file in the working directory.

Usage

crlrem( filename1, filename2 )

Arguments

Author(s)

This is code by Dirk Eddelbuettel copied from https://stat.ethz.ch/pipermail/r-devel/2010-September/058480.html

Examples

## Not run: 
filename1 <- "rm.arraymult__0.02.cpp"
filename2 <- "rm.arraymult__0.03.cpp"
crlrem( filename1, filename2 )
## End(Not run)

Description

Copies the Rcpp function into the working directory.

Usage

cxxfunction.copy(cppfct, name)

Arguments

References

Eddelbuettel, D. & Francois, R. (2011). Rcpp: Seamless R and C++ integration. Journal of Statistical Software, 40(8), 1-18. doi:10.18637/jss.v040.i08

See Also

inline::cxxfunction

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Rcpp code logistic distribution
#############################################################################

library(Rcpp)
library(inline)

# define Rcpp file
code1 <- "
    // input array A
    Rcpp::NumericMatrix AA(A);
    // Rcpp::IntegerVector dimAA(dimA);
    int nrows=AA.nrow();
    int ncolumns=AA.ncol();
    Rcpp::NumericMatrix Alogis(nrows,ncolumns)  ;
    // compute logistic distribution
    for (int ii=0; ii<nrows; ii++){
        Rcpp::NumericVector h1=AA.row(ii) ;
        Rcpp::NumericVector res=plogis( h1 ) ;
        for (int jj=0;jj<ncolumns;jj++){
            Alogis(ii,jj)=res[jj] ;
                        }
                    }
    return( wrap(Alogis) );
    "
# compile Rcpp code
fct_rcpp <- inline::cxxfunction( signature( A="matrix"), code1,
              plugin="Rcpp", verbose=TRUE )
# copy function and save it as object 'calclogis'
name <- "calclogis"  # name of the function
cxxfunction.copy( cppfct=fct_rcpp, name=name )
# function is available as object named as name
Reval( paste0( name, " <- fct_rcpp " ) )
# test function
m1 <- outer( seq( -2, 2, len=10 ), seq(-1.5,1.5,len=4) )
calclogis(m1)
    
## End(Not run)

Description

Datasets from Allison's missing data book (Allison 2002).

Usage

data(data.allison.gssexp)
data(data.allison.hip)
data(data.allison.usnews)

Format

• Data data.allison.gssexp:

  'data.frame': 2991 obs. of 14 variables:
$ AGE : num 33 59 NA 59 21 22 40 25 41 45 ...
$ EDUC : num 12 12 12 8 13 15 9 12 12 12 ...
$ FEMALE : num 1 0 1 0 1 1 1 0 1 1 ...
$ SPANKING: num 1 1 2 2 NA 1 3 1 1 NA ...
$ INCOM : num 11.2 NA 16.2 18.8 13.8 ...
$ NOCHILD : num 0 0 0 0 1 1 0 0 0 0 ...
$ NODOUBT : num NA NA NA 1 NA NA 1 NA NA 1 ...
$ NEVMAR : num 0 0 0 0 1 1 0 1 0 0 ...
$ DIVSEP : num 1 0 0 0 0 0 0 0 0 1 ...
$ WIDOW : num 0 0 0 0 0 0 1 0 1 0 ...
$ BLACK : num 1 1 1 0 1 1 0 1 1 1 ...
$ EAST : num 1 1 1 1 1 1 1 1 1 1 ...
$ MIDWEST : num 0 0 0 0 0 0 0 0 0 0 ...
$ SOUTH : num 0 0 0 0 0 0 0 0 0 0 ...
  

• Data data.allison.hip:

  'data.frame': 880 obs. of 7 variables:
$ SID : num 1 1 1 1 2 2 2 2 9 9 ...
$ WAVE: num 1 2 3 4 1 2 3 4 1 2 ...
$ ADL : num 3 2 3 3 3 1 2 1 3 3 ...
$ PAIN: num 0 5 0 0 0 1 5 NA 0 NA ...
$ SRH : num 2 4 2 2 4 1 1 2 2 3 ...
$ WALK: num 1 0 0 0 0 0 0 0 1 NA ...
$ CESD: num 9 28 31 11.6 NA ...
  

• Data data.allison.usnews:

  'data.frame': 1302 obs. of 7 variables:
$ CSAT : num 972 961 NA 881 NA ...
$ ACT : num 20 22 NA 20 17 20 21 NA 24 26 ...
$ STUFAC : num 11.9 10 9.5 13.7 14.3 32.8 18.9 18.7 16.7 14 ...
$ GRADRAT: num 15 NA 39 NA 40 55 51 15 69 72 ...
$ RMBRD : num 4.12 3.59 4.76 5.12 2.55 ...
$ PRIVATE: num 1 0 0 0 0 1 0 0 0 1 ...
$ LENROLL: num 4.01 6.83 4.49 7.06 6.89 ...
  

Source

The datasets were downloaded from http://www.ats.ucla.edu/stat/examples/md/.

References

Allison, P. D. (2002). Missing data. Newbury Park, CA: Sage.

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Hip dataset | Imputation using a wide format
#############################################################################

# at first, the hip dataset is 'melted' for imputation

data(data.allison.hip)
  ##   head(data.allison.hip)
  ##     SID WAVE ADL PAIN SRH WALK   CESD
  ##   1   1    1   3    0   2    1  9.000
  ##   2   1    2   2    5   4    0 28.000
  ##   3   1    3   3    0   2    0 31.000
  ##   4   1    4   3    0   2    0 11.579
  ##   5   2    1   3    0   4    0     NA
  ##   6   2    2   1    1   1    0  2.222

library(reshape)
hip.wide <- reshape::reshape(data.allison.hip, idvar="SID", timevar="WAVE",
                direction="wide")
  ##   > head(hip.wide, 2)
  ##     SID ADL.1 PAIN.1 SRH.1 WALK.1 CESD.1 ADL.2 PAIN.2 SRH.2 WALK.2 CESD.2 ADL.3
  ##   1   1     3      0     2      1      9     2      5     4      0 28.000     3
  ##   5   2     3      0     4      0     NA     1      1     1      0  2.222     2
  ##     PAIN.3 SRH.3 WALK.3 CESD.3 ADL.4 PAIN.4 SRH.4 WALK.4 CESD.4
  ##   1      0     2      0     31     3      0     2      0 11.579
  ##   5      5     1      0     12     1     NA     2      0     NA

# imputation of the hip wide dataset
imp <- mice::mice( as.matrix( hip.wide[,-1] ), m=5, maxit=3 )
summary(imp)

## End(Not run)

Description

Datasets from Enders' missing data book (2010).

Usage

data(data.enders.depression)
data(data.enders.eatingattitudes)
data(data.enders.employee)

Format

• Dataset data.enders.depression:

  'data.frame': 280 obs. of 8 variables:
$ txgroup: int 0 0 0 0 0 0 0 0 0 0 ...
$ dep1 : int 46 49 40 47 33 44 45 53 40 55 ...
$ dep2 : int 44 42 28 47 33 41 43 35 43 45 ...
$ dep3 : int 26 29 31 NA 34 34 34 35 35 36 ...
$ r2 : int 0 0 0 0 0 0 0 0 0 0 ...
$ r3 : int 0 0 0 1 0 0 0 0 0 0 ...
$ pattern: int 3 3 3 2 3 3 3 3 3 3 ...
$ dropout: int 0 0 0 1 0 0 0 0 0 0 ...
  

• Dataset data.enders.eatingattitudes:

  'data.frame': 400 obs. of 14 variables:
$ id : num 1 2 3 4 5 6 7 8 9 10 ...
$ eat1 : num 4 6 3 3 3 4 5 4 4 6 ...
$ eat2 : num 4 5 3 3 2 5 4 3 7 5 ...
$ eat10: num 4 6 2 4 3 4 4 4 6 5 ...
$ eat11: num 4 6 2 3 3 5 4 4 5 5 ...
$ eat12: num 4 6 3 4 3 4 4 4 4 6 ...
$ eat14: num 4 7 2 4 3 4 4 4 6 6 ...
$ eat24: num 3 6 3 3 3 4 4 4 4 5 ...
$ eat3 : num 4 5 3 3 4 4 3 6 4 5 ...
$ eat18: num 5 6 3 5 4 5 3 6 4 6 ...
$ eat21: num 4 5 2 4 4 4 3 5 4 5 ...
$ bmi : num 18.9 26 18.3 18.2 24.4 ...
$ wsb : num 9 13 6 5 10 7 11 8 10 12 ...
$ anx : num 11 19 8 14 7 11 12 12 14 12 ..
  

• Dataset data.enders.employee:

  'data.frame': 480 obs. of 9 variables:
$ id : num 1 2 3 4 5 6 7 8 9 10 ...
$ age : num 40 53 46 37 44 39 33 43 35 37 ...
$ tenure : num 10 14 10 8 9 10 7 9 9 10 ...
$ female : num 1 1 1 1 1 1 1 1 1 1 ...
$ wbeing : num 8 6 NA 7 NA 7 NA 7 7 5 ...
$ jobsat : num 8 5 7 NA 5 NA 5 NA 7 6 ...
$ jobperf : num 6 5 7 5 5 7 7 7 7 6 ...
$ turnover: num 0 0 0 0 0 0 0 0 1 0 ...
$ iq : num 106 93 107 94 107 118 103 106 108 97 ...
  

Source

The datasets were downloaded from https://www.appliedmissingdata.com/.

References

Enders, C. K. (2010). Applied missing data analysis. Guilford Press.

Description

Datasets from Grahams missing data book (2012).

Usage

data(data.graham.ex3)
data(data.graham.ex6)
data(data.graham.ex8a)
data(data.graham.ex8b)
data(data.graham.ex8c)

Format

• Dataset data.graham.ex3:

  'data.frame': 2756 obs. of 20 variables:
$ school : int 1 1 1 1 1 1 1 1 1 1 ...
$ alc7 : int 1 1 1 7 3 6 1 5 4 3 ...
$ rskreb71: int 1 3 1 2 1 NA 1 2 1 2 ...
$ rskreb72: int NA NA NA NA NA NA NA 3 2 3 ...
$ rskreb73: int NA NA NA NA NA NA NA 2 1 2 ...
$ rskreb74: int NA NA NA NA NA NA NA 3 2 4 ...
$ likepa71: int 4 2 3 3 2 NA 1 4 3 3 ...
$ likepa72: int 5 2 4 2 2 NA 5 3 3 2 ...
$ likepa73: int 4 1 3 3 2 NA 1 3 2 3 ...
$ likepa74: int 5 3 1 5 4 4 3 4 3 2 ...
$ likepa75: int 4 4 4 4 3 3 4 4 3 3 ...
$ posatt71: int 1 1 1 1 1 2 1 NA NA NA ...
$ posatt72: int 1 2 1 1 1 2 4 NA NA NA ...
$ posatt73: int 1 1 1 1 1 2 1 NA NA NA ...
$ alc8 : int 1 8 4 8 5 7 1 3 5 3 ...
$ rskreb81: int 1 4 1 2 2 3 2 3 1 4 ...
$ rskreb82: int NA NA NA NA NA NA NA 3 1 4 ...
$ rskreb83: int NA NA NA NA NA NA NA 2 1 2 ...
$ rskreb84: int NA NA NA NA NA NA NA 3 2 4 ...
$ alc9 : int 3 NA 7 NA 5 7 NA 6 6 7 ...
  

• Dataset data.graham.ex6:

  'data.frame': 2756 obs. of 9 variables:
$ school : int 1 1 1 1 1 1 1 1 1 1 ...
$ program : int 0 0 0 0 0 0 0 0 0 0 ...
$ alc7 : int 1 1 1 7 3 6 1 5 4 3 ...
$ riskreb7: int 1 3 1 2 1 NA 1 2 1 2 ...
$ likepar7: int 4 2 3 3 2 NA 1 4 3 3 ...
$ posatt7 : int 1 1 1 1 1 2 1 NA NA NA ...
$ alc8 : int 1 8 4 8 5 7 1 3 5 3 ...
$ riskreb8: int 1 4 1 2 2 3 2 3 1 4 ...
$ alc9 : int 3 NA 7 NA 5 7 NA 6 6 7 ...
  

• Dataset data.graham.ex8a:

  'data.frame': 1023 obs. of 20 variables:
$ skill1 : int 28 29 27 29 29 NA NA NA 29 NA ...
$ skill2 : int NA NA 29 29 NA NA NA NA NA 21 ...
$ skill3 : int NA NA 29 29 29 NA 28 10 29 25 ...
$ skill4 : int NA 29 25 29 29 28 29 NA NA NA ...
$ skill5 : int 29 29 28 28 29 NA 29 10 NA 25 ...
$ iplanV1: int 14 18 15 17 16 NA NA NA 18 NA ...
$ iplanV2: int NA NA 17 16 NA NA NA NA NA 16 ...
$ iplanV3: int NA NA 16 18 18 NA 17 1 18 16 ...
$ iplanV4: int NA 18 14 18 14 6 18 NA NA NA ...
$ iplanV5: int 13 18 12 18 18 NA 18 3 NA 5 ...
$ planA1 : int 1 0 2 8 3 NA NA NA 7 NA ...
$ planA2 : int NA NA 0 4 NA NA NA NA NA 6 ...
$ planA3 : int NA NA 1 4 7 NA 2 0 1 7 ...
$ planA4 : int NA 8 0 4 6 0 0 NA NA NA ...
$ planA5 : int 0 7 1 5 7 NA 2 0 NA 6 ...
$ planV1 : int NA NA NA NA NA NA NA NA NA NA ...
$ planV2 : int NA NA NA NA NA NA NA NA NA 1 ...
$ planV3 : int NA NA 1 NA NA NA NA 0 NA 1 ...
$ planV4 : int NA NA NA NA 2 NA NA NA NA NA ...
$ planV5 : int 2 NA 2 NA NA NA NA 0 NA NA ...
  

• Dataset data.graham.ex8b:

  'data.frame': 2570 obs. of 6 variables:
$ rskreb71: int 1 3 1 2 1 NA 1 2 1 2 ...
$ rskreb72: int NA NA NA NA NA NA NA 3 2 3 ...
$ posatt71: int 1 1 1 1 1 2 1 NA NA NA ...
$ posatt72: int 1 2 1 1 1 2 4 NA NA NA ...
$ posatt73: int 1 1 1 1 1 2 1 NA NA NA ...
$ posatt : int 3 4 3 3 3 6 6 NA NA NA ...
  

• Dataset data.graham.ex8c:

  'data.frame': 2756 obs. of 16 variables:
$ s1 : int 1 1 1 1 1 1 1 1 1 1 ...
$ s2 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s3 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s4 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s5 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s6 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s7 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s8 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s9 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s10 : int 0 0 0 0 0 0 0 0 0 0 ...
$ s11 : int 0 0 0 0 0 0 0 0 0 0 ...
$ xalc7 : int 1 1 1 7 3 6 1 5 4 3 ...
$ rskreb72: int NA NA NA NA NA NA NA 3 2 3 ...
$ likepa71: int 4 2 3 3 2 NA 1 4 3 3 ...
$ posatt71: int 1 1 1 1 1 2 1 NA NA NA ...
$ alc8 : int 1 8 4 8 5 7 1 3 5 3 ...
  

Source

The datasets were downloaded from http://methodology.psu.edu/pubs/books/missing.

References

Graham, J. W. (2012). Missing data. New York: Springer. doi:10.1007/978-1-4614-4018-5

Examples

## Not run: 
library(mitools)
library(mice)
library(Amelia)
library(jomo)

#############################################################################
# EXAMPLE 1: data.graham.8a | Imputation under multivariate normal model
#############################################################################

data(data.graham.ex8a)
dat <- data.graham.ex8a
dat <- dat[,1:10]
vars <- colnames(dat)
V <- length(vars)
# remove persons with completely missing data
dat <- dat[ rowMeans( is.na(dat) ) < 1, ]
summary(dat)

# some descriptive statistics
psych::describe(dat)

#**************
# imputation under a multivariate normal model
M <- 7  # number of imputations

#--------- mice package
# define imputation method
impM <- rep("norm", V)
names(impM) <- vars
# mice imputation
imp1a <- mice::mice( dat, method=impM, m=M, maxit=4 )
summary(imp1a)
# convert into a list of datasets
datlist1a <- miceadds::mids2datlist(imp1a)

#--------- Amelia package
imp1b <- Amelia::amelia( dat, m=M )
summary(imp1b)
datlist1b <- imp1b$imputations

#--------- jomo package
imp1c <- jomo::jomo1con(Y=dat, nburn=100, nbetween=10, nimp=M)
str(imp1c)
# convert into a list of datasets
datlist1c <- miceadds::jomo2datlist(imp1c)

# alternatively one can use the jomo wrapper function
imp1c1 <- jomo::jomo(Y=dat, nburn=100, nbetween=10, nimp=M)

#############################################################################
# EXAMPLE 2: data.graham.8b | Imputation with categorical variables
#############################################################################

data(data.graham.ex8b)
dat <- data.graham.ex8b
vars <- colnames(dat)
V <- length(vars)

# descriptive statistics
psych::describe(dat)

#*******************************
# imputation in mice using predictive mean matching
imp1a <- mice::mice( dat, m=5, maxit=10)
datlist1a <- mitools::imputationList( miceadds::mids2datlist(imp1a) )
print(datlist1a)

#*******************************
# imputation in jomo treating all variables as categorical

# Note that variables must have values from 1 to N
# use categorize function from sirt package here
dat.categ <- sirt::categorize( dat, categorical=colnames(dat), lowest=1 )
dat0 <- dat.categ$data

# imputation in jomo treating all variables as categorical
Y_numcat <- apply( dat0, 2, max, na.rm=TRUE )
imp1b <- jomo::jomo1cat(Y.cat=dat0, Y.numcat=Y_numcat, nburn=100,
                 nbetween=10, nimp=5)

# recode original categories
datlist1b <- sirt::decategorize( imp1b, categ_design=dat.categ$categ_design )
# convert into a list of datasets
datlist1b <- miceadds::jomo2datlist(datlist1b)
datlist1b <- mitools::imputationList( datlist1b )

# Alternatively, jomo can be used but categorical variables must be
# declared as factors
dat <- dat0
# define two variables as factors
vars <- miceadds::scan.vec(" rskreb71 rskreb72")
for (vv in vars){
    dat[, vv] <- as.factor( dat[,vv] )
          }
# use jomo
imp1b1 <- jomo::jomo(Y=dat, nburn=30, nbetween=10, nimp=5)

#****************************
# compare frequency tables for both imputation packages
fun_prop <- function( variable ){
            t1 <- table(variable)
            t1 / sum(t1)
                }

# variable rskreb71
res1a <-  with( datlist1a, fun_prop(rskreb71) )
res1b <-  with( datlist1b, fun_prop(rskreb71) )
summary( miceadds::NMIcombine(qhat=res1a, NMI=FALSE ) )
summary( miceadds::NMIcombine(qhat=res1b, NMI=FALSE ) )

# variable posatt
res2a <-  with( datlist1a, fun_prop(posatt) )
res2b <-  with( datlist1b, fun_prop(posatt) )
summary( miceadds::NMIcombine(qhat=res2a, NMI=FALSE ) )
summary( miceadds::NMIcombine(qhat=res2b, NMI=FALSE ) )

## End(Not run)

Description

Dataset with items corresponding to internet attitudes.

Usage

data(data.internet)

Format

A data frame with 281 observations on the following 22 variables.

The format of the dataset is

'data.frame': 281 obs. of 22 variables:
$ IN1 : num 1 5 2 3 1 3 2 3 2 1 ...
$ IN2 : num 4 3 2 7 7 4 4 7 4 3 ...
$ IN3 : num 4 5 4 2 1 2 5 2 2 4 ...
[...]
$ IN20: num 3 2 2 3 3 4 2 7 2 2 ...
$ IN21: num 3 3 6 5 4 4 5 5 6 5 ...
$ IN22: num 3 4 2 5 3 5 3 7 3 5 ...

Details

The following text is copied from http://people.few.eur.nl/groenen/Data/index.htm

The data set is based on a questionnaire on attitudes towards the Internet. It consists of evaluations of 22 statements about the Internet by 281 students at Erasmus University Rotterdam. These data were gathered around 2002 before the wide availability of broadband Internet access in the Netherlands. The statements were evaluated using a seven-point Likert scale, ranging from 1 (completely disagree) to 7 (completely agree).

We would like to thank Peter Verhoef for making these data available.

Each variable (statement) is coded as follows:

1. Completely disagree
2. Disagree
3. Slightly disagree
4. Neutral
5. Slightly agree
6. Agree
7. Completely agree

Internet items:

1. Paying using Internet is safe
2. Surfing the Internet is easy
3. Internet is unreliable
4. Internet is slow
5. Internet is user-friendly
6. Internet is the future's means of communication
7. Internet is addictive
8. Internet is fast
9. Sending personal data using the Internet is unsafe
10. The prices of Internet subscriptions are high
11. Internet offers many possibilities for abuse
12. The costs of surfing are high
13. Internet offers unbounded opportunities
14. Internet phone costs are high
15. The content of web sites should be regulated
16. Internet is easy to use
17. I like surfing
18. I often speak with friends about the Internet
19. I like to be informed of important new things
20. I always attempt new things on the Internet first
21. I regularly visit websites recommended by others
22. I know much about the Internet

Source

Peter Verhoef

http://people.few.eur.nl/groenen/Data/index.htm

Examples

data(data.internet)
# missing proportions
colMeans( is.na(data.internet) )

Description

Large-scale dataset with many cases and few variables included for testing purposes.

Usage

data(data.largescale)

Format

A data frame with 14000 observations on the following 13 variables. The format is

'data.frame': 14000 obs. of 13 variables:
$ id: num 1e+07 1e+07 1e+07 1e+07 1e+07 ...
$ D1: num 0 0 0 0 1 0 0 0 0 0 ...
$ D2: num 0 0 0 1 0 1 0 1 0 0 ...
$ D3: num 0 0 0 0 0 0 0 0 0 0 ...
$ D4: num 0 0 0 1 0 0 0 1 0 0 ...
$ D5: num 0 0 0 0 0 1 0 0 0 0 ...
$ v1: num 118 117 94 106 86 117 96 96 82 95 ...
$ v2: num 101 101 86 101 65 94 72 75 70 99 ...
$ v3: num 0 0 0 0 0 1 0 0 0 0 ...
$ v4: num 3 NA 3 5 2 5 5 5 4 2 ...
$ v5: num 0 NA 0 0 0 1 0 0 0 0 ...
$ v6: num 3 3 3 4 NA 1 3 3 2 3 ...
$ v7: num 51 36 14 47 22 17 13 37 47 38 ...

Description

Example datasets for miceadds package.

Usage

data(data.ma01)
data(data.ma02)
data(data.ma03)
data(data.ma04)
data(data.ma05)
data(data.ma06)
data(data.ma07)
data(data.ma08)

Format

• Dataset data.ma01:

  Dataset with students nested within school and student weights (studwgt). The format is

  'data.frame': 4073 obs. of 11 variables:
$ idstud : num 1e+07 1e+07 1e+07 1e+07 1e+07 ...
$ idschool: num 1001 1001 1001 1001 1001 ...
$ studwgt : num 6.05 6.05 5.27 5.27 6.05 ...
$ math : int 594 605 616 524 685 387 536 594 387 562 ...
$ read : int 647 651 539 551 689 502 503 597 580 576 ...
$ migrant : int 0 0 0 1 0 0 1 0 0 0 ...
$ books : int 6 6 5 2 6 3 4 6 6 5 ...
$ hisei : int NA 77 69 45 66 53 43 NA 64 50 ...
$ paredu : int 3 7 7 2 7 3 4 NA 7 3 ...
$ female : int 1 1 0 0 1 1 0 0 1 1 ...
$ urban : num 1 1 1 1 1 1 1 1 1 1 ...
  

• Dataset data.ma02:

  10 multiply imputed datasets of incomplete data data.ma01. The format is

  List of 10
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
$ :'data.frame': 4073 obs. of 11 variables:
  

• Dataset data.ma03:

  This dataset contains one variable math_EAP for which a conditional posterior distribution with EAP and its associated standard deviation is available.

  'data.frame': 120 obs. of 8 variables:
$ idstud : int 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 ...
$ female : int 0 1 1 1 1 0 1 1 1 1 ...
$ migrant : int 1 1 0 1 1 0 0 0 1 0 ...
$ hisei : int 44 NA 26 NA 32 60 31 NA 34 26 ...
$ educ : int NA 2 NA 1 4 NA 2 NA 2 NA ...
$ read_wle : num 74.8 78.1 103.2 81.2 119.2 ...
$ math_EAP : num 337 342 264 285 420 ...
$ math_SEEAP: num 28 29.5 28.6 28.5 27.5 ...
  

• Dataset data.ma04:

  This dataset contains two hypothetical scales A and B and single variables V5, V6 and V7.

  'data.frame': 281 obs. of 13 variables:
$ group: int 1 1 1 1 1 1 1 1 1 1 ...
$ A1 : int 2 2 2 1 1 3 3 NA 2 1 ...
$ A2 : int 2 2 2 3 1 2 4 4 4 4 ...
$ A3 : int 2 3 3 4 1 3 2 2 2 4 ...
$ A4 : int 3 4 6 4 7 5 3 5 5 1 ...
$ V5 : int 2 2 5 5 4 3 4 1 3 4 ...
$ V6 : int 2 5 5 1 1 3 2 2 2 4 ...
$ V7 : int 6 NA 4 5 6 2 5 5 6 7 ...
$ B1 : int 7 NA 6 4 5 2 5 7 3 7 ...
$ B2 : int 6 NA NA 6 3 3 4 6 6 7 ...
$ B3 : int 7 NA 7 4 3 4 3 7 5 NA ...
$ B4 : int 4 5 6 5 4 3 4 5 2 1 ...
$ B5 : int 7 NA 7 4 4 3 5 7 5 4 ...
  

• Dataset data.ma05:

  This is a two-level dataset with students nested within classes. Variables at the student level are Dscore, Mscore, denote, manote, misei and migrant. Variables at the class level are sprengel and groesse.

  'data.frame': 1673 obs. of 10 variables:
$ idstud : int 100110001 100110002 100110003 100110004 100110005 ...
$ idclass : int 1001 1001 1001 1001 1001 1001 1001 1001 1001 1001 ...
$ Dscore : int NA 558 643 611 518 552 NA 534 409 543 ...
$ Mscore : int 404 563 569 621 653 651 510 NA 517 566 ...
$ denote : int NA 1 1 1 3 2 3 2 3 2 ...
$ manote : int NA 1 1 1 1 1 2 2 2 1 ...
$ misei : int NA 51 NA 38 NA 50 53 53 38 NA ...
$ migrant : int NA 0 0 NA 0 0 0 0 0 NA ...
$ sprengel: int 0 0 0 0 0 0 0 0 0 0 ...
$ groesse : int 25 25 25 25 25 25 25 25 25 25 ...
  

• Dataset data.ma06:

  This is a dataset in which the variable FC is only available with grouped values (coarse data or interval data).

  'data.frame': 198 obs. of 7 variables:
$ id : num 1001 1002 1003 1004 1005 ...
$ A1 : int 14 7 10 15 0 5 9 6 8 0 ...
$ A2 : int 5 6 4 8 2 5 4 0 7 0 ...
$ Edu : int 4 3 1 5 5 1 NA 1 5 3 ...
$ FC : int 3 2 2 2 2 NA NA 2 2 NA ...
$ FC_low: num 10 5 5 5 5 0 0 5 5 0 ...
$ FC_upp: num 15 10 10 10 10 100 100 10 10 100 ...
  

• Dataset data.ma07:

  This is a three-level dataset in which the variable FC is only available with grouped values (coarse data or interval data).

  'data.frame': 1600 obs. of 9 variables:
$ id3: num 1001 1001 1001 1001 1001 ...
$ id2: num 101 101 101 101 101 101 101 101 101 101 ...
$ id1: int 1 2 3 4 5 6 7 8 9 10 ...
$ x1 : num 0.91 1.88 NA 1.52 0.93 0.51 2.11 0.99 2.42 NA ...
$ x2 : num -0.58 1.12 0.87 -0.01 -0.14 0.48 1.85 -0.9 0.93 0.63 ...
$ y1 : num 1.66 1.66 1.66 1.66 1.66 1.66 1.66 1.66 1.66 1.66 ...
$ y2 : num 0.96 0.96 0.96 0.96 0.96 0.96 0.96 0.96 0.96 0.96 ...
$ z1 : num -0.53 -0.53 -0.53 -0.53 -0.53 -0.53 -0.53 -0.53 -0.53 -0.53 ...
$ z2 : num 0.42 0.42 0.42 0.42 0.42 0.42 0.42 0.42 0.42 0.42 ...
  

• Dataset data.ma08:

  List with several vector of strings containing descriptive data from published articles. See string_to_matrix for converting these strings into matrices.

  List of 4
$ mat1: chr [1:6] "1. T1_mental_health" ...
$ mat2: chr [1:16] "1. Exp voc-T1 -" ...
$ mat3: chr [1:12] "1. TOWRE age 7\t-\t\t\t\t\t\t" ...
$ mat4: chr [1:18] "1. Vocab. age 7\t-\t\t\t\t\t" ...
  

• Dataset data.ma09:

  This is a subset of a PISA dataset that is used for generating synthetic data.

  'data.frame': 342 obs. of 41 variables:
$ SEX : int 1 2 1 2 1 2 2 2 2 1 ...
$ AGE : num 16 15.9 16.3 15.5 15.9 ...
$ HISEI : int 37 46 66 51 25 NA 54 52 51 69 ...
$ FISCED : int 3 3 6 3 3 NA 3 3 2 2 ...
$ MISCED : int 3 4 4 4 3 NA 4 3 4 4 ...
$ PV1MATH: num 643 556 510 604 462 ...
$ M474Q01: int 1 1 1 1 0 1 1 1 1 0 ...
$ M155Q02: int 2 2 2 2 2 0 0 2 2 2 ...
$ M155Q01: int 1 1 0 1 1 1 1 1 1 1 ...
[...]
  

Description

Small-scale dataset for testing purposes (moderate number of cases, many variables)

Usage

data(data.smallscale)

Format

A data frame with 675 observations on the following 164 variables. The format is

'data.frame': 675 obs. of 164 variables:
$ v1 : num 3 3 2 3 3 0 1 0 3 NA ...
$ v2 : num 3 0 1 3 0 0 0 3 2 NA ...
$ v3 : num 0 0 2 3 2 0 1 0 0 NA ...
$ v4 : num 1 3 3 3 NA 0 0 0 3 NA ...
$ v5 : num 0 0 3 3 0 0 3 1 3 3 ...
$ v6 : num 8 8 9 8 9 9 9 8 9 9 ...
[...]

Description

Creates objects of class datlist or nested.datlist.

The functions nested.datlist2datlist and datlist2nested.datlist provide list conversions for imputed datasets.

Usage

datlist_create(datasets)

nested.datlist_create(datasets)

## S3 method for class 'datlist'
print(x, ...)

## S3 method for class 'nested.datlist'
print(x, ...)

nested.datlist2datlist(datlist)

datlist2nested.datlist(datlist, Nimp)

Arguments

Value

Object of class datlist or nested.datlist

Examples

## Not run: 
## The function datlist_create is currently defined as
function (datasets)
{
    class(datasets) <- "datlist"
    return(datasets)
  }

#############################################################################
# EXAMPLE 1: Create object of class datlist
#############################################################################

library(BIFIEsurvey)
data(data.timss2, package="BIFIEsurvey" )
datlist <- data.timss2

# class datlist
obj1 <- miceadds::datlist_create(data.timss2)

#############################################################################
# EXAMPLE 2: Multiply imputed datasets: Different object classes
#############################################################################

library(mice)
data(nhanes2, package="mice")
set.seed(990)

# nhanes2 data imputation
imp1 <- miceadds::mice.1chain( nhanes2, burnin=5, iter=25, Nimp=5 )
# object of class datlist
imp2 <- miceadds::mids2datlist(imp1)
# alternatively, one can use datlist_create
imp2b <- miceadds::datlist_create(imp1)
# object of class imputationList
imp3 <- mitools::imputationList(imp2)
# retransform object in class datlist
imp2c <- miceadds::datlist_create(imp3)
str(imp2c)

#############################################################################
# EXAMPLE 3: Nested multiply imputed datasets
#############################################################################

library(BIFIEsurvey)
data(data.timss2, package="BIFIEsurvey" )
datlist <- data.timss2
   # list of 5 datasets containing 5 plausible values

#** define imputation method and predictor matrix
data <- datlist[[1]]
V <- ncol(data)
# variables
vars <- colnames(data)
# variables not used for imputation
vars_unused <- miceadds::scan.vec("IDSTUD TOTWGT  JKZONE  JKREP" )

#- define imputation method
impMethod <- rep("norm", V )
names(impMethod) <- vars
impMethod[ vars_unused ] <- ""

#- define predictor matrix
predM <- matrix( 1, V, V )
colnames(predM) <- rownames(predM) <- vars
diag(predM) <- 0
predM[, vars_unused ] <- 0

# object of class nmi
imp1 <- miceadds::mice.nmi( datlist, method=impMethod, predictorMatrix=predM,
                m=4, maxit=3 )
# object of class nested.datlist
imp2 <- miceadds::mids2datlist(imp1)
# object of class NestedImputationList
imp3 <- miceadds::NestedImputationList(imp2)
# redefine class nested.datlist
imp4 <- miceadds::nested.datlist_create(imp3)

#############################################################################
# EXAMPLE 4: Conversions between nested lists of datasets and lists of datasets
#############################################################################

library(BIFIEsurvey)
data(data.timss4, package="BIFIEsurvey" )
datlist <- data.timss4

# object of class nested.datlist
datlist1a <- miceadds::nested.datlist_create(datlist)
# object of class NestedImputationList
datlist1b <- miceadds::NestedImputationList(datlist)

# conversion to datlist
datlist2a <- miceadds::nested.datlist2datlist(datlist1a)  # class datlist
datlist2b <- miceadds::nested.datlist2datlist(datlist1b)  # class imputationList

# convert into a nested list with 2 between nests and 10 within nests
datlist3a <- miceadds::datlist2nested.datlist(datlist2a, Nimp=c(2,10) )
datlist3b <- miceadds::datlist2nested.datlist(datlist2b, Nimp=c(4,5) )

## End(Not run)

Description

This function converts a list of multiply imputed data sets to an object of class amelia.

Usage

datlist2Amelia(datlist)

Arguments

Value

An object of class amelia

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Imputation of NHANES data using mice package
#############################################################################

library(mice)
library(Amelia)

data(nhanes,package="mice")
set.seed(566)  # fix random seed

# imputation with mice
imp <- mice::mice(nhanes, m=7)

# conversion to amelia object
amp <- miceadds::datlist2Amelia(datlist=imp)
str(amp)
plot(amp)
print(amp)
summary(amp)

## End(Not run)

Description

This function converts a list of multiply imputed data sets to a mice::mids object.

Usage

datlist2mids(dat.list, progress=FALSE)
datalist2mids(dat.list, progress=FALSE)

Arguments

Value

An object of class mids

See Also

See mice::as.mids for converting a multiply imputed dataset in long format into a mids object.

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Imputation of NHANES data using Amelia package
#############################################################################

library(mice)
library(Amelia)

data(nhanes,package="mice")
set.seed(566)  # fix random seed

# impute 10 datasets using Amelia
a.out <- Amelia::amelia(x=nhanes, m=10)
# plot of observed and imputed data
plot(a.out)

# convert list of multiply imputed datasets into a mids object
a.mids <- miceadds::datlist2mids( a.out$imputations )

# linear regression: apply mice functionality lm.mids
mod <- with( a.mids, stats::lm( bmi ~ age ) )
summary( mice::pool( mod ) )
  ##                    est       se         t       df     Pr(>|t|)     lo 95
  ##  (Intercept) 30.624652 2.626886 11.658158 8.406608 1.767631e-06 24.617664
  ##  age         -2.280607 1.323355 -1.723352 8.917910 1.192288e-01 -5.278451
  ##                   hi 95 nmis       fmi    lambda
  ##  (Intercept) 36.6316392   NA 0.5791956 0.4897257
  ##  age          0.7172368    0 0.5549945 0.4652567

# fit linear regression model in Zelig
library(Zelig)
mod2 <- Zelig::zelig( bmi ~ age, model="ls", data=a.out, cite=FALSE)
summary(mod2)
  ##  Model: Combined Imputations
  ##              Estimate Std.Error z value Pr(>|z|)
  ##  (Intercept)   30.625     2.627  11.658  0.00000 ***
  ##  age           -2.281     1.323  -1.723  0.08482
  ##  ---
  ##  Signif. codes:  '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

# fit linear regression using mitools package
library(mitools)
datimp <- mitools::imputationList(a.out$imputations)
mod3 <- with( datimp, stats::lm( bmi ~ age ) )
summary( mitools::MIcombine( mod3 ) )
  ##  Multiple imputation results:
  ##        with(datimp, stats::lm(bmi ~ age))
  ##        MIcombine.default(mod3)
  ##                results       se    (lower     upper) missInfo
  ##  (Intercept) 30.624652 2.626886 25.304594 35.9447092     51 
  ##  age         -2.280607 1.323355 -4.952051  0.3908368     49 

## End(Not run)

Description

This function provides unidimensional plausible value imputation with a known measurement error variance or classical test theory (Mislevy, 1991). The reliability of the scale is estimated by Cronbach's Alpha or can be provided by the user.

Usage

draw.pv.ctt(y, dat.scale=NULL, x=NULL, samp.pars=TRUE,
      alpha=NULL, sig.e=NULL, var.e=NULL, true.var=NULL)

Arguments

Details

The linear model is assumed for drawing plausible values of a variable $Y$ contaminated by measurement error. Assuming $Y=\theta + e$ and a linear regression model for $\theta$

$\theta=\bold{X} \beta + \epsilon$

(plausible value) imputations from the posterior distribution $P( \theta | Y, \bold{X} )$ are drawn. See Mislevy (1991) for details.

Value

A vector with plausible values

Note

Plausible value imputation is also labeled as multiple overimputation (Blackwell, Honaker & King, 2011).

References

Blackwell, M., Honaker, J., & King, G. (2011). Multiple overimputation: A unified approach to measurement error and missing data. Technical Report.

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

See Also

See also sirt::plausible.value.imputation.raschtype for plausible value imputation.

Plausible value imputations can be conducted in mice using the imputation method mice.impute.plausible.values.

Plausible values can be drawn in Amelia by specifying observation-level priors, see Amelia::moPrep and Amelia::amelia.

Examples

## Not run: 

#############################################################################
# SIMULATED EXAMPLE 1: Scale scores
#############################################################################

set.seed(899)
n <- 5000       # number of students
x <- round( stats::runif( n, 0,1 ) )
y <- stats::rnorm(n)
# simulate true score theta
theta <- .6 + .4*x + .5 * y + stats::rnorm(n)
# simulate observed score by adding measurement error
sig.e <- rep( sqrt(.40), n )
theta_obs <- theta + stats::rnorm( n, sd=sig.e)

# calculate alpha
( alpha <- stats::var( theta ) / stats::var( theta_obs ) )
# [1] 0.7424108
#=> Ordinarily, sig.e or alpha will be known, assumed or estimated by using items,
#    replications or an appropriate measurement model.

# create matrix of predictors
X <- as.matrix( cbind(x, y ) )

# plausible value imputation with scale score
imp1 <- miceadds::draw.pv.ctt( y=theta_obs, x=X, sig.e=sig.e )
# check results
stats::lm( imp1 ~ x + y )

# imputation with alpha as an input
imp2 <- miceadds::draw.pv.ctt( y=theta_obs, x=X, alpha=.74 )
stats::lm( imp2 ~ x + y )

#--- plausible value imputation in Amelia package
library(Amelia)

# define data frame
dat <- data.frame( "x"=x, "y"=y, "theta"=theta_obs )
# generate observation-level priors for theta
priors <- cbind( 1:n, 3, theta_obs, sig.e )
             # 3 indicates column index for theta
overimp <- priors[,1:2]
# run Amelia
imp <- Amelia::amelia( dat, priors=priors, overimp=overimp, m=10)
# create object of class datlist and evaluate results
datlist <- miceadds::datlist_create( imp$imputations )
withPool_MI( with( datlist, stats::var(theta) ) )
stats::var(theta)       # compare with true variance
mod <- with( datlist,  stats::lm( theta ~ x + y ) )
mitools::MIcombine(mod)

## End(Not run)

Description

The function filename_split splits a file name into parts.

The function string_extract_part extracts a part of a string.

The function string_to_matrix converts a string into a matrix.

Usage

filename_split(file_name, file_sep="__", file_ext=".")
filename_split_vec( file_names, file_sep="__", file_ext=".")

string_extract_part( vec, part=1, sep="__", remove_empty=TRUE )

string_to_matrix(x, rownames=NULL, col_elim=NULL, as_numeric=FALSE,
               diag_val=NULL, extend=FALSE, col1_numeric=FALSE, split=" ")

Arguments

Value

List with components of the file name (see Examples).

See Also

files_move

Examples

#############################################################################
# EXAMPLE 1: Demonstration example for filename_split
#############################################################################

# file name
file_name <- "pisa_all_waves_invariant_items_DATA_ITEMS_RENAMED__DESCRIPTIVES__2016-10-12_1000.csv"

# apply function
miceadds::filename_split( file_name )
  ##  $file_name
  ##  [1] "pisa_all_waves_invariant_items_DATA_ITEMS_RENAMED__DESCRIPTIVES__2016-10-12_1000.csv"
  ##  $stem
  ##  [1] "pisa_all_waves_invariant_items_DATA_ITEMS_RENAMED__DESCRIPTIVES"
  ##  $suffix
  ##  [1] "2016-10-12_1000"
  ##  $ext
  ##  [1] "csv"
  ##  $main
  ##  [1] "pisa_all_waves_invariant_items_DATA_ITEMS_RENAMED__DESCRIPTIVES.csv"

#############################################################################
# EXAMPLE 2: Example string_extract_part
#############################################################################

vec <- c("ertu__DES", "ztu__DATA", "guzeuue745_ghshgk34__INFO", "zzu78347834_ghghwuz")

miceadds::string_extract_part( vec=vec, part=1, sep="__" )
miceadds::string_extract_part( vec=vec, part=2, sep="__" )
  ##  > miceadds::string_extract_part( vec=vec, part=1, sep="__" )
  ##  [1] "ertu"                "ztu"                 "guzeuue745_ghshgk34"
  ##  [4] "zzu78347834_ghghwuz"
  ##  > miceadds::string_extract_part( vec=vec, part=2, sep="__" )
  ##  [1] "DES"  "DATA" "INFO" NA

## Not run: 
#############################################################################
# EXAMPLE 3: Reading descriptive information from published articles
#############################################################################
data(data.ma08)
library(stringr)

#**** reading correlations (I)
dat <- data.ma08$mat1
miceadds::string_to_matrix(dat, rownames=2, col_elim=c(1,2))

#**** reading correlations including some processing (II)
dat0 <- data.ma08$mat2
dat <- dat0[1:14]

# substitute "*"
dat <- gsub("*", "", dat, fixed=TRUE )

# replace blanks in variable names
s1 <- stringr::str_locate(dat, "[A-z] [A-z]")
start <- s1[,"start"] + 1
for (ss in 1:length(start) ){
    if ( ! is.na( start[ss] ) ){
        substring( dat[ss], start[ss], start[ss] ) <- "_"
    }
}

# character matrix
miceadds::string_to_matrix(dat)
# numeric matrix containing correlations
miceadds::string_to_matrix(dat, rownames=2, col_elim=c(1,2), as_numeric=TRUE, diag_val=1,
           extend=TRUE )
#** reading means and SDs
miceadds::string_to_matrix(dat0[ c(15,16)], rownames=1, col_elim=c(1), as_numeric=TRUE )

#**** reading correlations (III)
dat <- data.ma08$mat3
dat <- gsub(" age ", "_age_", dat, fixed=TRUE )
miceadds::string_to_matrix(dat, rownames=2, col_elim=c(1,2), as_numeric=TRUE, diag_val=1,
       extend=TRUE )

#**** reading correlations (IV)
dat <- data.ma08$mat4 <- dat0

# remove spaces in variable names
dat <- gsub(" age ", "_age_", dat, fixed=TRUE )
s1 <- stringr::str_locate_all(dat, "[A-z,.] [A-z]")
NL <- length(dat)
for (ss in 1:NL ){
    NR <- nrow(s1[[ss]])
    if (NR>1){
        start <- s1[[ss]][2,1]+1
        if ( ! is.na( start ) ){
            substring( dat[ss], start, start ) <- "_"
        }
    }
}

miceadds::string_to_matrix(dat, rownames=2, col_elim=c(1,2), as_numeric=TRUE, diag_val=1,
     extend=TRUE )

#############################################################################
# EXAMPLE 4: Input string of length one
#############################################################################


pm0 <- "
0.828
0.567 0.658
0.664 0.560 0.772
0.532 0.428 0.501 0.606
0.718 0.567 0.672 0.526 0.843"

miceadds::string_to_matrix(x=pm0, as_numeric=TRUE, extend=TRUE)

#############################################################################
# EXAMPLE 5: String with variable names and blanks
#############################################################################

tab1 <- "
Geometric Shapes .629 .021 (.483) -.049 (.472)
Plates .473 .017 (.370) .105 (.405)
Two Characteristics .601 .013 (.452) -.033 (.444)
Crossing Out Boxes .597 -.062 (.425) -.036 (.445)
Numbers/Letters .731 .004 (.564) .003 (.513)
Numbers/Letters mixed .682 .085 (.555) .082 (.514)"

miceadds::string_to_matrix(x=tab1, col1_numeric=TRUE)

## End(Not run)

Description

Moves older (defined in alphanumeric order) files from one directory to another directory. If directories do not exist, they will be automatically created.

Usage

files_move(path1, path2, file_sep="__", pattern=NULL, path2_name="__ARCH")

Arguments

See Also

filename_split

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Move older files in '__ARCHIVE' directory
#############################################################################

# specify path
path1 <- "p:/IPN/Projects/PISA/Trend_2015/2__Data/All_Waves/"
# specify target directory which is an archive
path2 <- file.path( path1, "__ARCHIVE" )
# move files
files_move( path1, path2 )

## End(Not run)

Description

Simulates univariate non-normal data by using Fleishman power transformations (Fleishman, 1978; Demirtas & Hedeker, 2007).

Usage

fleishman_sim(N=1, coef=NULL, mean=0, sd=1, skew=0, kurt=0)

fleishman_coef(mean=0, sd=1, skew=0, kurt=0)

Arguments

Details

For $Z \sim N(0,1)$, the Fleishman power normal variable $X$ is defined as $X=a + bZ + cZ^2 + d Z^3$.

Value

Vector of simulated values (fleishman_sim) or list of coefficients (fleishman_coef).

References

Demirtas, H., & Hedeker, D. (2008). Imputing continuous data under some non-Gaussian distributions. Statistica Neerlandica, 62(2), 193-205. doi:10.1111/j.1467-9574.2007.00377.x

Fleishman, A. I. (1978). A method for simulating non-normal distributions. Psychometrika, 43(4), 521-532. doi:10.1007/BF02293811

See Also

See also the BinOrdNonNor::Fleishman.coef.NN function in the BinOrdNonNor package.

See the nnig_sim function for simulating a non-normally distributed multivariate variables.

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Simulating values with Fleishman polynomial
#############################################################################

#* define mean, standard deviation, skewness and kurtosis
mean <- .75
sd <- 2
skew <- 1
kurt <- 3

#* compute coefficients of Fleishman polynomial
coeff <- miceadds::fleishman_coef(mean=mean, sd=sd, skew=skew, kurt=kurt)
print(coeff)

# sample size
N <- 1000
set.seed(2018)
#* simulate values based on estimated coefficients
X <- miceadds::fleishman_sim(N=N, coef=coeff)
#* simulate values based on input of moments
X <- miceadds::fleishman_sim(N=N, mean=mean, sd=sd, skew=skew, kurt=kurt)

## End(Not run)

Description

These functions slightly extend the usage of grep but it is extended to a vector argument.

Usage

grep.vec(pattern.vec, x, operator="AND", ...)

grepvec( pattern.vec, x, operator="AND", value=FALSE, ...)

grep_leading( pattern, x, value=FALSE )

grepvec_leading( patternvec, x, value=FALSE )

Arguments

Examples

#############################################################################
# EXAMPLE 1: Toy example
#############################################################################

vec <- c("abcd", "bcde", "aedf", "cdf" )
# search for entries in vec with contain 'a' and 'f'
#  -> operator="AND"
grep.vec( pattern.vec=c("a","f"), x=vec )
  ##   $x
  ##   [1] "aedf"
  ##   $index.x
  ##   [1] 3

grepvec( pattern.vec=c("a","f"), x=vec, value=TRUE)
grepvec( pattern.vec=c("a","f"), x=vec, value=FALSE)

# search for entries in vec which contain 'a' or 'f'
grep.vec( pattern.vec=c("a","f"), x=vec, operator="OR")
  ##   $x
  ##   [1] "abcd" "aedf" "cdf"
  ##   $index.x
  ##   [1] 1 3 4

Description

Calculates some groupwise descriptive statistics.

Usage

GroupMean(data, group, weights=NULL, extend=FALSE, elim=FALSE)

GroupSum(data, group, weights=NULL, extend=FALSE)

GroupSD(data, group, weights=NULL, extend=FALSE)

# group mean of a variable
gm(y, cluster, elim=FALSE)

# centering within clusters
cwc(y, cluster)

Arguments

Value

A data frame or a vector with groupwise calculated statistics

See Also

mitml::clusterMeans

base::rowsum, stats::aggregate, stats::ave

Examples

## Not run: 

#############################################################################
# EXAMPLE 1: Group means and standard deviations for data.ma02
#############################################################################

data(data.ma02, package="miceadds" )
dat <- data.ma02[[1]] # select first dataset

#--- group means for read and math
GroupMean( dat[, c("read","math") ], group=dat$idschool )
# using rowsum
a1 <- base::rowsum( dat[, c("read","math") ], dat$idschool )
a2 <- base::rowsum( 1+0*dat[, c("read","math") ], dat$idschool )
(a1/a2)[1:10,]
# using aggregate
stats::aggregate(  dat[, c("read","math") ], list(dat$idschool), mean )[1:10,]

#--- extend group means to original dataset
GroupMean( dat[, c("read","math") ], group=dat$idschool, extend=TRUE )
# using ave
stats::ave( dat[, "read" ], dat$idschool  )
stats::ave( dat[, "read" ], dat$idschool, FUN=mean )

#--- group standard deviations
GroupSD( dat[, c("read","math") ], group=dat$idschool)[1:10,]
# using aggregate
stats::aggregate(  dat[, c("read","math") ], list(dat$idschool), sd )[1:10,]

#############################################################################
# EXAMPLE 2: Calculating group means and group mean centering
#############################################################################

data(data.ma07, package="miceadds")
dat <- data.ma07

# compute group means
miceadds::gm( dat$x1, dat$id2 )
# centering within clusters
miceadds::cwc( dat$x1, dat$id2 )

# evaluate formula with model.matrix
X <- model.matrix( ~ I( miceadds::cwc(x1, id2) ) + I( miceadds::gm(x1,id2) ), data=dat )
head(X)

## End(Not run)

Description

Indicator function for analyzing coverage. The output indicates whether a value lies within a computed confidence interval.

Usage

in_CI(est, se, true, level=0.95, df=Inf)

Arguments

Value

Logical vector

Examples

#############################################################################
# EXAMPLE 1: Toy example
#############################################################################

#-- simulate estimates and standard errors
set.seed(987)
n <- 10
est <- stats::rnorm( n, sd=1)
se <- stats::runif( n, 0, .7 )
level <- .95
true <- 0

#-- apply coverage function
in_ci <- miceadds::in_CI( est, se, true)
#-- check correctness
cbind( est, se, true, in_ci )

Description

This function includes an index variable to a data frame in the first column.

Usage

index.dataframe(data,systime=FALSE)

Arguments

Examples

dfr <- matrix( 2*1:12-3, 4,3 )
colnames(dfr) <- paste0("X",1:ncol(dfr))
index.dataframe( dfr)
  ##     index X1 X2 X3
  ##   1     1 -1  7 15
  ##   2     2  1  9 17
  ##   3     3  3 11 19
  ##   4     4  5 13 21
index.dataframe( dfr, systime=TRUE)
  ##     index         file_created X1 X2 X3
  ##   1     1  2013-08-22 10:26:28 -1  7 15
  ##   2     2  2013-08-22 10:26:28  1  9 17
  ##   3     3  2013-08-22 10:26:28  3 11 19
  ##   4     4  2013-08-22 10:26:28  5 13 21

Description

Converts a jomo data frame in long format into a list of datasets or an object of class mids.

Usage

jomo2datlist(jomo.dataframe, variable="Imputation")

jomo2mids(jomo.dataframe, variable="Imputation")

Arguments

Value

List of multiply imputed datasets

See Also

See the jomo package.

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Dataset nhanes | jomo imputation and conversion into a data list
#############################################################################

data(nhanes, package="mice")
dat <- nhanes

# impute under multivariate normal model in jomo
imp1 <- jomo::jomo1con(Y=dat, nburn=100, nbetween=10, nimp=5)
# convert into a list of datasets
datlist1 <- miceadds::jomo2datlist(imp1)
# convert into mids object
datlist2 <- miceadds::jomo2datlist(imp1)

## End(Not run)

Description

Fits a PLS regression model with the kernel algorithm (Dayal & Macgregor, 1997).

Usage

kernelpls.fit2(X, Y, ncomp)

## S3 method for class 'kernelpls.fit2'
predict(object,X, ...)

Arguments

Value

The same list as in {pls::kernelpls.fit} is produced.

In addition, $R^2$ measures are contained in R2.

Author(s)

This code is a Rcpp translation of the original pls::kernelpls.fit function from the pls package (see Mevik & Wehrens, 2007).

References

Dayal, B., & Macgregor, J. F. (1997). Improved PLS algorithms. Journal of Chemometrics, 11(1), 73-85.

Mevik, B. H., & Wehrens, R. (2007). The pls package: Principal component and partial least squares regression in R. Journal of Statistical Software, 18, 1-24. doi:10.18637/jss.v018.i02

See Also

See the pls package for further estimation algorithms.

Examples

## Not run: 
#############################################################################
# SIMULATED EXAMPLE 1: 300 cases on 100 variables
#############################################################################
set.seed(789)
library(mvtnorm)

N <- 300        # number of cases
p <- 100        # number of predictors
rho1 <- .6      # correlations between predictors

# simulate data
Sigma <- base::diag(1-rho1,p) + rho1
X <- mvtnorm::rmvnorm( N, sigma=Sigma )
beta <- base::seq( 0, 1, len=p )
y <- ( X %*% beta )[,1] + stats::rnorm( N, sd=.6 )
Y <- base::matrix(y,nrow=N, ncol=1 )

# PLS regression
res <- miceadds::kernelpls.fit2( X=X, Y=Y, ncomp=20 )

# predict new scores
Xpred <- predict( res, X=X[1:10,] )

#############################################################################
# EXAMPLE 2: Dataset yarn from pls package
#############################################################################

# use kernelpls.fit from pls package
library(pls)
data(yarn,package="pls")
mod1 <- pls::kernelpls.fit( X=yarn$NIR, Y=yarn$density, ncomp=10 )
# use kernelpls.fit2 from miceadds package
Y <- base::matrix( yarn$density, ncol=1 )
mod2 <- miceadds::kernelpls.fit2( X=yarn$NIR, Y=Y, ncomp=10 )

## End(Not run)

Description

Loads packages specified in vector pkg. If some packages are not yet installed, they will be automatically installed by this function using install.packages.

Usage

library_install( pkg, ... )

Arguments

Examples

## Not run: 
# try to load packages PP and MCMCglmm
library_install( pkg=c("PP", "MCMCglmm") )

## End(Not run)

Description

Computes cluster robust standard errors for linear models (stats::lm) and general linear models (stats::glm) using the multiwayvcov::vcovCL function in the sandwich package.

Usage

lm.cluster(data, formula, cluster, weights=NULL, subset=NULL )

glm.cluster(data, formula, cluster, weights=NULL, subset=NULL, family="gaussian" )

## S3 method for class 'lm.cluster'
summary(object,...)
## S3 method for class 'glm.cluster'
summary(object,...)

## S3 method for class 'lm.cluster'
coef(object,...)
## S3 method for class 'glm.cluster'
coef(object,...)

## S3 method for class 'lm.cluster'
vcov(object,...)
## S3 method for class 'glm.cluster'
vcov(object,...)

Arguments

Value

List with following entries

Note

If lm.cluster is used inside a function, add wgt__ <<- weight for assigning the weight to wgt__ in the global environment.

See Also

stats::lm, stats::glm, sandwich::vcovCL

Examples

## Not run: 

#############################################################################
# EXAMPLE 1: Cluster robust standard errors data.ma01
#############################################################################

data(data.ma01)
dat <- data.ma01

#*** Model 1: Linear regression
mod1 <- miceadds::lm.cluster( data=dat, formula=read ~ hisei + female,
               cluster="idschool" )
coef(mod1)
vcov(mod1)
summary(mod1)

# estimate Model 1, but cluster is provided as a vector
mod1b <- miceadds::lm.cluster( data=dat, formula=read ~ hisei + female,
                 cluster=dat$idschool)
summary(mod1b)

#*** Model 2: Logistic regression
dat$highmath <- 1 * ( dat$math > 600 )   # create dummy variable
mod2 <- miceadds::glm.cluster( data=dat, formula=highmath ~ hisei + female,
                cluster="idschool", family="binomial")
coef(mod2)
vcov(mod2)
summary(mod2)

#############################################################################
# EXAMPLE 2: Cluster robust standard errors for multiply imputed datasets
#############################################################################

library(mitools)
data(data.ma05)
dat <- data.ma05

# imputation of the dataset: use six imputations
resp <- dat[, - c(1:2) ]
imp <- mice::mice( resp, method="norm", maxit=3, m=6 )
datlist <- miceadds::mids2datlist( imp )

# linear regression with cluster robust standard errors
mod <- lapply(  datlist, FUN=function(data){
            miceadds::lm.cluster( data=data, formula=denote ~ migrant+ misei,
                    cluster=dat$idclass )
            }  )
# extract parameters and covariance matrix
betas <- lapply( mod, FUN=function(rr){ coef(rr) } )
vars <- lapply( mod, FUN=function(rr){ vcov(rr) } )
# conduct statistical inference
summary( miceadds::pool_mi( qhat=betas, u=vars ) )

#------ compute global F-test for hypothesis that all predictors have zero coefficient values
library(mitml)
Nimp <- 6 # number of imputations
np <- length(betas[[1]])   # number of parameters
beta_names <- names(betas[[1]])
# define vector of parameters for which constraints should be tested
constraints <- beta_names[-1]
# create input for mitml::testConstraints function
qhat <- matrix( unlist(betas), ncol=Nimp)
rownames(qhat) <- beta_names
uhat <- array( unlist(vars), dim=c(np,np,Nimp))
dimnames(uhat) <- list( beta_names, beta_names, NULL )
# compute global F-test
Ftest <- mitml::testConstraints( qhat=betas, uhat=vars, constraints=constraints )
print(Ftest)

#############################################################################
# EXAMPLE 3: Comparing miceadds::lm.cluster() and lme4::lmer()
#############################################################################

data(data.ma01, package="miceadds")
dat <- na.omit(data.ma01)

# center hisei variable
dat$hisei <- dat$hisei - mean(dat$hisei)

# define school mean hisei
dat$hisei_gm <- miceadds::GroupMean(dat$hisei, dat$idschool, extend=TRUE)[,2]
dat$cluster_size <- miceadds::GroupSum(1+0*dat$hisei, dat$idschool, extend=TRUE)[,2]
dat$hisei_wc <- dat$hisei - dat$hisei_gm

#*** Model 1a: lm, hisei with clustering
mod1a <- miceadds::lm.cluster( data=dat, formula=read~hisei, cluster="idschool" )

#*** Model 1b: lmer, hisei
mod1b <- lme4::lmer( data=dat, formula=read~hisei+(1|idschool) )

cbind( coef(mod1a), fixef(mod1b))
 ##  > cbind( coef(mod1a), fixef(mod1b))
 ##                    [,1]        [,2]
 ##  (Intercept) 509.181691 507.8684752
 ##  hisei         1.524776   0.8161745

# variance explanation
vmod1b <- r2mlm::r2mlm(mod1b)
vmod1b$Decompositions

#*** Model 2a: lm, hisei and hisei_gm with clustering
mod2a <- miceadds::lm.cluster( data=dat, formula=read~hisei_wc+hisei_gm,
                                   cluster="idschool" )

#*** Model 2b: lmer, multilevel model
mod2b <- lme4::lmer( data=dat, formula=read~hisei_wc+hisei_gm + (1|idschool) )

# variance explanation
vmod2b <- r2mlm::r2mlm(mod2b)
vmod2b$Decompositions

cbind( coef(mod2a), fixef(mod2b))
 ##  > cbind( coef(mod2a), fixef(mod2b))
 ##                     [,1]        [,2]
 ##  (Intercept) 509.1816911 508.0478629
 ##  hisei_wc      0.7503773   0.7503773
 ##  hisei_gm      5.8424012   5.5681941


## End(Not run)

Description

The function lmer_vcov conducts statistical inference for fixed coefficients and standard deviations and correlations of random effects structure of models fitted in the lme4 package.

The function lmer_pool applies the Rubin formula for inference for fitted lme4 models for multiply imputed datasets.

Usage

lmer_vcov(object, level=.95, use_reml=FALSE, ...)

## S3 method for class 'lmer_vcov'
summary(object, digits=4, file=NULL, ...)
## S3 method for class 'lmer_vcov'
coef(object, ...)
## S3 method for class 'lmer_vcov'
vcov(object, ...)

lmer_vcov2(object, level=.95, ...)

lmer_pool( models, level=.95, ...)
## S3 method for class 'lmer_pool'
summary(object, digits=4, file=NULL, ...)

lmer_pool2( models, level=.95, ...)

Arguments

Value

List with several entries:

Author(s)

Function originally from Ben Bolker, http://rpubs.com/bbolker/varwald

See Also

lme4::lmer, mitml::testEstimates

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Single model fitted in lme4
#############################################################################

library(lme4)
data(data.ma01, package="miceadds")
dat <- na.omit(data.ma01)

#* fit multilevel model
formula <- math ~ hisei + miceadds::gm( books, idschool ) + ( 1 + books | idschool )
mod1 <- lme4::lmer( formula, data=dat, REML=FALSE)
summary(mod1)

#* statistical inference
res1 <- miceadds::lmer_vcov( mod1 )
summary(res1)
coef(res1)
vcov(res1)

#############################################################################
# EXAMPLE 2: lme4 model for multiply imputed dataset
#############################################################################

library(lme4)
data(data.ma02, package="miceadds")
datlist <- miceadds::datlist_create(data.ma02)

#** fit lme4 model for all imputed datasets
formula <- math ~ hisei + miceadds::gm( books, idschool ) + ( 1 | idschool )
models <- list()
M <- length(datlist)
for (mm in 1:M){
    models[[mm]] <- lme4::lmer( formula, data=datlist[[mm]], REML=FALSE)
}

#** statistical inference
res1 <- miceadds::lmer_pool(models)
summary(res1)

## End(Not run)

Description

The function load.data is a wrapper function for loading or reading data frames or matrices.

The function load.files loads multiple files in a data frame.

Usage

load.data( filename, type=NULL, path=getwd(), load_fun=NULL, spss.default=TRUE, ...)

load.files( files, type=NULL, path=getwd(), ...)

Arguments

See Also

See also load.Rdata for loading R data frames.

See save.Rdata and save.data for saving/writing R data frames.

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Toy example
#############################################################################

# load a data frame in the file "data_s3.Rdata" and save this
# as the object "dat.s3"
dat.s3 <- miceadds::load.data( filename="data_s3.Rdata", type="Rdata" )
print(str(dat.s3))

# load text input with base::readLines() function using the 'load_fun' argument
dat <- miceadds::load.data( "my_output_", type="Rout", load_fun=readLines, path=path)

## End(Not run)

Description

These functions loads a Rdata object saved as a data frame or a matrix in the current R environment. The function load.Rdata saves the loaded object in the global environment while load.Rdata2 loads the object only specified environments. Hence, usage of load.Rdata2 instead of load.Rdata is recommended.

Usage

load.Rdata(filename, objname)

load.Rdata2(filename, path=getwd(), RDS=FALSE)

Arguments

See Also

See also save.Rdata for saving data frames in a Rdata format.

See also: base::load, base::save

Examples

## Not run: 
# load a data frame in the file "data_s3.Rdata" and save this
# as the object "dat.s3"
load.Rdata( filename="data_s3.Rdata", "dat.s3" )
head(dat.s3)

# Alternatively one can use the function
dat.s3 <- miceadds::load.Rdata2( filename="data_s3.Rdata")

## End(Not run)

Description

Utility functions for working with lme4 formula objects. The function ma_lme4_formula_terms decomposes an lme4 formula into several parts for further processing.

Usage

ma_lme4_formula_terms(formula)

ma_lme4_formula_design_matrices(formula, data, start_index=0, formula_terms=NULL,
        only_design_matrices=FALSE)

Arguments

Value

List with several entries

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Splitting a lme4 formula
#############################################################################

#*** formula for a multilevel model
formula <- y ~ I( miceadds::cwc(x, idcluster)) + z + I(z^2) + I( miceadds::gm(x, idcluster) ) + w +
                        ( x + I(x^2) | idcluster)  + (0 +  w | idcluster ) +
                        ( 0 + I(as.factor(f)) | idcluster)
miceadds::ma_lme4_formula_terms(formula)

#*** formula for a single level model
formula2 <- y ~ I( miceadds::cwc(x, idcluster)) + z + I(z^2) + I( miceadds::gm(x, idcluster) ) + w
miceadds::ma_lme4_formula_terms(formula2)

#############################################################################
# EXAMPLE 2: Design matrices for multilevel model
#############################################################################

data(data.ma07, package="miceadds")
dat <- data.ma07

formula <- x1 ~ x2 + I( miceadds::gm( x2, id2)) + I( miceadds::gm( x2, id3)) + y1 + z1 +
                    ( x2 | id2:id3 ) + ( 1 | id3 ) + ( 0 + x2 | id3 )
res <- miceadds::ma_lme4_formula_design_matrices(formula, data=dat)
str(res)

## End(Not run)

Description

Some functions for normally distributed data.

The function ma_rmvnorm is like mvtnorm::rmvnorm, but allows for a covariance matrix sigma which can have zero variances.

Usage

ma_rmvnorm(n, mu=NULL, sigma, eps=1e-10)

Arguments

Value

Matrix of simulated values

See Also

MASS::mvrnorm

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Two-dimensional simulation with zero variance at dimension 1
#############################################################################

sigma <- matrix( c(0,0,0,1), nrow=2, ncol=2)
miceadds::ma_rmvnorm( n=10, sigma=sigma )

## End(Not run)

Description

This function performs a z-standardization for a numeric matrix. Note that in a case of a zero standard deviation all matrix entries are divided by a small number such that no NaNs occur.

Usage

ma.scale2(x, missings=FALSE)

Arguments

Value

A matrix

See Also

base::scale

Examples

#############################################################################
# EXAMPLE 1: z-standardization data.internet
#############################################################################

data(data.internet)
dat <- data.internet

# z-standardize all variables in this dataset
zdat <- miceadds::ma.scale2( dat, missings=TRUE )

## Not run: 
#############################################################################
# SIMULATED EXAMPLE 2: Speed comparison for many cases and many variables
#############################################################################

set.seed(9786)
# 3000 cases, 200 variables
N <- 3000
p <- 200
# simulate some data
x <- matrix( stats::rnorm( N*p ), N, p )
x <- round( x, 2 )

# compare computation times for 10 replications
B <- 10
    s1 <- Sys.time()        # scale in R
for (bb in 1:B){
    res <- scale(x)
} ; s2 <- Sys.time() ; d1 <- s2-s1

    s1 <- Sys.time()        # scale in miceadds
for (bb in 1:B){
    res1 <- miceadds::ma.scale2(x)
} ; s2 <- Sys.time() ; d2 <- s2-s1

# scale in miceadds with missing handling
    s1 <- Sys.time()
for (bb in 1:B){
    res1 <- miceadds::ma.scale2(x,missings=TRUE)
} ; s2 <- Sys.time() ; d3 <- s2-s1
d1      # scale in R
d2      # scale in miceadds (no missing handling)
d3      # scale in miceadds (with missing handling)
  ##   > d1      # scale in R
  ##   Time difference of 1.622431 secs
  ##   > d2      # scale in miceadds (no missing handling)
  ##   Time difference of 0.156003 secs
  ##   > d3      # scale in miceadds (with missing handling)
  ##   Time difference of 0.2028039 secs

## End(Not run)

Description

Some multivariate descriptive statistics for weighted datasets in miceadds. A list of (nested) multiply imputed data sets is also allowed as input.

Usage

ma.wtd.meanNA(data, weights=NULL, vars=NULL )

ma.wtd.sdNA(data, weights=NULL, vars=NULL, method="unbiased" )

ma.wtd.covNA(data, weights=NULL, vars=NULL, method="unbiased" )

ma.wtd.corNA(data, weights=NULL, vars=NULL, method="unbiased" )

ma.wtd.skewnessNA(data, weights=NULL, vars=NULL, method="unbiased" )

ma.wtd.kurtosisNA(data, weights=NULL, vars=NULL, method="unbiased" )

ma.wtd.quantileNA( data, weights=NULL, vars=NULL, type=7,
          probs=seq(0,1,.25) )

Arguments

Details

Contrary to ordinary R practice, missing values are ignored in the calculation of descriptive statistics.

Value

A vector or a matrix depending on the requested statistic.

Note

If data is of class BIFIEdata and no weights are specified, sample weights are extracted from the BIFIEdata object.

See Also

Some functions for weighted statistics: stats::weighted.mean, stats::cov.wt, {Hmisc::wtd.var}, TAM::weighted_quantile, ...

See micombine.cor for statistical inference of correlation coefficients.

Examples

#############################################################################
# EXAMPLE 1: Weighted statistics for a single dataset data.ma01
#############################################################################

data(data.ma01)
dat <- as.matrix(data.ma01[,-c(1:3)])

# weighted mean
ma.wtd.meanNA( dat, weights=data.ma01$studwgt )

# weighted SD
ma.wtd.sdNA( dat, weights=data.ma01$studwgt )

# weighted covariance for selected variables
ma.wtd.covNA( dat, weights=data.ma01$studwgt, vars=c("books","hisei") )

# weighted correlation
ma.wtd.corNA( dat, weights=data.ma01$studwgt )

## Not run: 
# weighted skewness
ma.wtd.skewnessNA( dat[,"books"], weights=data.ma01$studwgt )
# compare with result in TAM
TAM::weighted_skewness( x=dat[,"books"], w=data.ma01$studwgt )

# weighted kurtosis
ma.wtd.kurtosisNA( dat, weights=data.ma01$studwgt, vars=c("books","hisei") )
# compare with TAM
TAM::weighted_kurtosis( dat[,"books"], w=data.ma01$studwgt )
TAM::weighted_kurtosis( dat[,"hisei"], w=data.ma01$studwgt )

#############################################################################
# EXAMPLE 2: Weighted statistics multiply imputed dataset
#############################################################################

library(mitools)
data(data.ma05)
dat <- data.ma05

# do imputations
resp <- dat[, - c(1:2) ]
# object of class mids
imp <- mice::mice( resp, method="norm", maxit=3, m=5 )
# object of class datlist
datlist <- miceadds::mids2datlist( imp )
# object of class imputationList
implist <- mitools::imputationList(datlist)

# weighted means
ma.wtd.meanNA(datlist)
ma.wtd.meanNA(implist)
ma.wtd.meanNA(imp)

# weighted quantiles
ma.wtd.quantileNA( implist, weights=data.ma05$studwgt, vars=c("manote","Dscore"))

#############################################################################
# EXAMPLE 3: Weighted statistics nested multiply imputed dataset
#############################################################################

library(BIFIEsurvey)
data(data.timss2, package="BIFIEsurvey" )
datlist <- data.timss2   # list of 5 datasets containing 5 plausible values

#** define imputation method and predictor matrix
data <- datlist[[1]]
V <- ncol(data)
# variables
vars <- colnames(data)
# variables not used for imputation
vars_unused <- miceadds::scan.vec("IDSTUD TOTWGT  JKZONE  JKREP" )
#- define imputation method
impMethod <- rep("norm", V )
names(impMethod) <- vars
impMethod[ vars_unused ] <- ""
#- define predictor matrix
predM <- matrix( 1, V, V )
colnames(predM) <- rownames(predM) <- vars
diag(predM) <- 0
predM[, vars_unused ] <- 0

# object of class mids.nmi
imp1 <- miceadds::mice.nmi( datlist, method=impMethod, predictorMatrix=predM,
                m=4, maxit=3 )
# object of class nested.datlist
datlist <- miceadds::mids2datlist(imp1)
# object of class NestedImputationList
imp2 <- miceadds::NestedImputationList(datlist)

# weighted correlations
vars <- c("books","ASMMAT","likesc")
ma.wtd.corNA( datlist,  vars=vars )
ma.wtd.corNA( imp2,  vars=vars )
ma.wtd.corNA( imp1,  vars=vars )

#############################################################################
# EXAMPLE 4: Multiply imputed datasets in BIFIEdata format
#############################################################################

library(BIFIEsurvey)
data(data.timss1, package="BIFIEsurvey")
data(data.timssrep, package="BIFIEsurvey")

# create BIFIEdata object
bdat <- BIFIEsurvey::BIFIE.data( data.list=data.timss1, wgt=data.timss1[[1]]$TOTWGT,
            wgtrep=data.timssrep[, -1 ] )
summary(bdat)
# create BIFIEdata object in a compact way
bdat2 <- BIFIEsurvey::BIFIE.data( data.list=data.timss1, wgt=data.timss1[[1]]$TOTWGT,
            wgtrep=data.timssrep[, -1 ], cdata=TRUE)
summary(bdat2)

# compute skewness
ma.wtd.skewnessNA( bdat, vars=c("ASMMAT", "books" ) )
ma.wtd.skewnessNA( bdat2, vars=c("ASMMAT", "books" ) )

#############################################################################
# EXAMPLE 5: Nested multiply imputed datasets in BIFIEdata format
#############################################################################

data(data.timss4, package="BIFIEsurvey")
data(data.timssrep, package="BIFIEsurvey")

# nested imputed dataset, save it in compact format
bdat <- BIFIE.data( data.list=data.timss4, wgt=data.timss4[[1]][[1]]$TOTWGT,
            wgtrep=data.timssrep[, -1 ], NMI=TRUE, cdata=TRUE )
summary(bdat)
# skewness
ma.wtd.skewnessNA( bdat, vars=c("ASMMAT", "books" ) )

## End(Not run)

Description

Computes Cohen's d effect size indicating whether missingness on a variable is related to other variables (covariates).

Usage

mi_dstat(dat)

Arguments

Value

A matrix. Missingness indicators refer to rows and covariates to columns.

Examples

#############################################################################
# EXAMPLE 1: d effect size for missingness indicators data.ma01
#############################################################################

data(data.ma01)
dat <- data.ma01

# compute d effect sizes
md <- miceadds::mi_dstat(dat)
round( md, 3 )

Description

This function combines $F$ values from analysis of variance using the $D_2$ statistic which is based on combining $\chi^2$ statistics (see Allison, 2001, Grund, Luedtke & Robitzsch, 2016; micombine.F, micombine.chisquare).

Usage

mi.anova(mi.res, formula, type=2)

Arguments

Value

A list with the following entries:

References

Allison, P. D. (2002). Missing data. Newbury Park, CA: Sage.

Grund, S., Luedtke, O., & Robitzsch, A. (2016). Pooling ANOVA results from multiply imputed datasets: A simulation study. Methodology, 12(3), 75-88. doi:10.1027/1614-2241/a000111

See Also

This function uses micombine.F and micombine.chisquare.

See mice::pool.compare and mitml::testModels for model comparisons based on the $D_1$ statistic. The $D_2$ statistic is also included in mitml::testConstraints.

The $D_1$, $D_2$ and $D_3$ statistics are also included in the mice package in functions mice::D1, mice::D2 and mice::D3.

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: nhanes2 data | two-way ANOVA
#############################################################################

library(mice)
library(car)
data(nhanes2, package="mice")
set.seed(9090)

# nhanes data in one chain and 8 imputed datasets
mi.res <- miceadds::mice.1chain( nhanes2, burnin=4, iter=20, Nimp=8 )
# 2-way analysis of variance (type 2)
an2a <- miceadds::mi.anova(mi.res=mi.res, formula="bmi ~ age * chl" )

# test of interaction effects using mitml::testModels()
mod1 <- with( mi.res, stats::lm( bmi ~ age*chl ) )
mod0 <- with( mi.res, stats::lm( bmi ~ age+chl ) )

mitml::testModels(model=mod1$analyses, null.model=mod0$analyses, method="D1")
mitml::testModels(model=mod1$analyses, null.model=mod0$analyses, method="D2")

# 2-way analysis of variance (type 3)
an2b <- miceadds::mi.anova(mi.res=mi.res, formula="bmi ~ age * chl", type=3)

#****** analysis based on first imputed dataset

# extract first dataset
dat1 <- mice::complete( mi.res$mids )

# type 2 ANOVA
lm1 <- stats::lm( bmi ~ age * chl, data=dat1 )
summary( stats::aov( lm1 ) )
# type 3 ANOVA
lm2 <- stats::lm( bmi ~ age * chl, data=dat1, contrasts=list(age=contr.sum))
car::Anova(mod=lm2, type=3)

## End(Not run)

Description

The function mice.impute.2l.continuous imputes values of continuous variables with a linear mixed effects model using lme4::lmer or blme::blmer. The lme4::lmer or blme::blmer function is also used for predictive mean matching where the match is based on predicted values which contain the fixed and (sampled) random effects. Binary variables can be imputed from a two-level logistic regression model fitted with the lme4::glmer or blme::bglmer function. See Snijders and Bosker (2012) and Zinn (2013) for details.

Usage

mice.impute.2l.continuous(y, ry, x, type, intercept=TRUE,
    groupcenter.slope=FALSE, draw.fixed=TRUE, random.effects.shrinkage=1E-6,
    glmer.warnings=TRUE, blme_use=FALSE, blme_args=NULL, ... )

mice.impute.2l.pmm(y, ry, x, type, intercept=TRUE,
    groupcenter.slope=FALSE, draw.fixed=TRUE, random.effects.shrinkage=1E-6,
    glmer.warnings=TRUE, donors=5, match_sampled_pars=TRUE,
    blme_use=FALSE, blme_args=NULL, ... )

mice.impute.2l.binary(y, ry, x, type, intercept=TRUE,
    groupcenter.slope=FALSE, draw.fixed=TRUE, random.effects.shrinkage=1E-6,
    glmer.warnings=TRUE, blme_use=FALSE, blme_args=NULL, ... )

Arguments

Value

A vector of length nmis=sum(!ry) with imputed values.

References

Snijders, T. A. B., & Bosker, R. J. (2012). Multilevel analysis: An introduction to basic and advanced multilevel modeling. Thousand Oaks, CA: Sage.

Zinn, S. (2013). An imputation model for multilevel binary data. NEPS Working Paper No 31.

See Also

See mice.impute.ml.lmer for imputation for datasets with more than two levels (e.g., three-level datasets or cross-classified datasets).

Variables at a higher level (e.g. at level 2) can be imputed using 2lonly functions, for example the mice::mice.impute.2lonly.norm function in the mice package or the general mice.impute.2lonly.function function in the miceadds package which using an already defined imputation method at level 1. If a level-2 variable for 3-level data should be imputed, then mice.impute.ml.lmer can also be used to impute this variable with a two-level imputation model in which level 1 corresponds to the original level-2 units and level 2 corresponds to the original level-3 units.

See mice::mice.impute.2l.norm and mice::mice.impute.2l.pan for imputation functions in the mice package under fully conditional specification for normally distributed variables. The function mice::mice.impute.2l.norm allows for residual variances which are allowed to vary across groups while mice::mice.impute.2l.pan assumes homogeneous residual variances.

The micemd package provides further imputation methods for the mice package for imputing multilevel data with fully conditional specification. The function micemd::mice.impute.2l.jomo has similar functionality like mice::mice.impute.2l.pan and imputes normally distributed two-level data with a Bayesian MCMC approach, but relies on the jomo package instead of the pan package. The functions mice::mice.impute.2l.lmer and micemd::mice.impute.2l.glm.norm have similar functionality like mice.impute.2l.continuous and imputes normally distributed two-level data. The function {micemd::mice.impute.2l.glm.bin} has similar functionality like mice.impute.2l.binary and imputes binary two-level data.

The hmi package imputes single-level and multilevel data and is also based on fully conditional specification. The package relies on the MCMC estimation implemented in the MCMCglmm package. The imputation procedure can be run with the hmi::hmi function.

See the pan (pan::pan) and the jomo (jomo::jomo) package for joint multilevel imputation. See mitml::panImpute and mitml::jomoImpute for wrapper functions to these packages in the mitml package.

Imputation by chained equations can also be conducted in blocks of multivariate conditional distributions since mice 3.0.0 (see the blocks argument in mice::mice). The mitml::panImpute and mitml::jomoImpute functions can be used with mice::mice by specifying imputation methods "panImpute" (see mice::mice.impute.panImpute)) and "jomoImpute" (see mice::mice.impute.jomoImpute)).

Examples

## Not run: 
#############################################################################
# EXAMPLE 1: Imputation of a binary variable
#############################################################################

#--- simulate missing values
set.seed(976)
G <- 30        # number of groups
n <- 8        # number of persons per group
iccx <- .2    # intra-class correlation X
iccy <- .3    # latent intra-class correlation binary outcome
bx <- .4    # regression coefficient
threshy <- stats::qnorm(.70)  # threshold for y
x <- rep( rnorm( G, sd=sqrt( iccx) ), each=n )  +
            rnorm(G*n, sd=sqrt( 1 - iccx) )
y <- bx * x + rep( rnorm( G, sd=sqrt( iccy) ), each=n )  +
                rnorm(G*n, sd=sqrt( 1 - iccy) )
y <- 1 * ( y > threshy )
dat <- data.frame( group=100+rep(1:G, each=n), x=x, y=y )

#* create some missings
dat1 <- dat
dat1[ seq( 1, G*n, 3 ),"y" ]  <- NA
dat1[ dat1$group==2, "y" ] <- NA

#--- prepare imputation in mice
vars <- colnames(dat1)
V <- length(vars)
#* predictor matrix
predmat <- matrix( 0, nrow=V, ncol=V)
rownames(predmat) <- colnames(predmat) <- vars
predmat["y", ] <- c(-2,2,0)
#* imputation methods
impmeth <- rep("",V)
names(impmeth) <- vars
impmeth["y"] <- "2l.binary"

#** imputation with logistic regression ('2l.binary')
imp1 <- mice::mice( data=as.matrix(dat1), method=impmeth,
                predictorMatrix=predmat, maxit=1, m=5 )

#** imputation with predictive mean matching ('2l.pmm')
impmeth["y"] <- "2l.pmm"
imp2 <- mice::mice( data=as.matrix(dat1), method=impmeth,
                predictorMatrix=predmat, maxit=1, m=5 )

#** imputation with logistic regression using blme package
blme_args <- list( "cov.prior"="invwishart")
imp3 <- mice::mice( data=as.matrix(dat1), method=impmeth,
                predictorMatrix=predmat, maxit=1, m=5,
                blme_use=TRUE, blme_args=blme_args )

## End(Not run)