README.md

In mmansolf/TSI: Custom mice Functions for True Score Imputation

TSI

True score imputation: Custom imputation function for mice in R

Credit line

This package includes code Adapted with permission from Wolters Kluwer Health, Inc.: Guo, Y., Little, R. J., & McConnell, D. S. (2012). On using summary statistics from an external calibration sample to correct for covariate measurement error. Epidemiology, 165-174. Full text link: https://journals.lww.com/epidem/Fulltext/2012/01000/On_Using_Summary_Statistics_From_an_External.25.aspx

Description

Generates data sets using true score imputation to account for psychometric measurement error. Imputation is conducted by a call to the mice function from the eponymous package, using method='truescore' to call the custom imputation function mice.impute.truescore included in this package. Reliability or standard error of measurement are incorporated through the blots argument to mice. Because these function calls can be complicated, a convenience function TSI is included which creates and executes the mice call for the user.

Example

The following code uses an included data set, data_eap, containing two variables containing EAP-generated T scores, Fx and Fy, and associated standard errors, SE.Fx and SE.Fy, respectively. This code conducts five imputations with five burn-in iterations per imputation, generating imputed true score variables Tx and Ty. More details on this example can be found in the vignette, obtained by running vignette('TSI') after installing the package.

library(TSI)
set.seed(0)
mice.data=TSI(data_eap,
              os_names=c('Fx','Fy'),
              se_names=c('SE.Fx','SE.Fy'),
              metrics='T',
              score_types='EAP',
              separated=T,
              ts_names=c('Tx','Ty'),
              mice_args=c(m=5,maxit=5,printFlag=F))
mice.data

The resulting mids object can be analyzed using convenience functions available for the mice package:

pool(with(mice.data,lm(Ty~Tx+m)))

Directly calling the mice package

The above code uses a convenience function to generate a call to mice to generate the imputations. Instead, one can call mice directly, although the function call is more complicated when constructed by hand. As above, see vignette('TSI') for more details.

library(TSI)
set.seed(0)
mice.data=mice(data_eap_2,m=5,maxit=5,
  method=c('pmm','pmm','pmm','pmm','pmm',
           'truescore','truescore'),
  blocks=list(Fx="Fx",Fy="Fy",SE.Fx="SE.Fx",SE.Fy="SE.Fy",m="m",
              Tx='Tx',Ty='Ty'),
  blots=list(Tx=list(calibration=list(os_name='Fx',
                                      se_name='SE.Fx',
                                      score_type='EAP',
                                      mean=50,
                                      var_ts=100,
                                      separated=T)),
             Ty=list(calibration=list(os_name='Fy',
                                      se_name="SE.Fy",
                                      score_type='EAP',
                                      mean=50,
                                      var_ts=100,
                                      separated=T))),
  predictorMatrix=matrix(c(0,1,1,1,1,0,0,
                           1,0,1,1,1,0,0,
                           1,1,0,1,1,0,0,
                           1,1,1,0,1,0,0,
                           1,1,1,1,0,0,0,
                           1,1,1,1,1,0,0,
                           1,1,1,1,1,0,0),7,7,byrow=T),
  printFlag=F,
  remove.constant=F)
mice.data

pool(with(mice.data,lm(Ty~Tx+m)))

Installing from source

We recommend installing the latest development version of this package from Github to ensure use of the latest and greatest version. To do so: 1. Install the devtools package if you haven't. In R, this can be done using the following code: install.packages('devtools') 2. Load the devtools package using the following code: library(devtools) 3. Install the TSI package using install_github() using the following code:

devtools::install_github('TSI',build_vignettes=T)

The ,build_vignettes=T is optional but recommended for viewing the vignette accompanying this package. The TSI package is not yet on CRAN, so trying to install it with install.packages() will not work. This is coming soon!

Other installation options

While the above method should work for most users, there are alternatives: The ghit library has an analogous ghit::install_github() function to that in the devtools package Download the package as a .zip file, then run the following code and interactively select the .zip file to install: install.packages(file.choose(), repos = NULL, type = "source")

Additional license information

LICENSE: Creative Commons - Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) For more information on this license type, see https://creativecommons.org/licenses/by-nc/4.0/ For more information on licensing for this package in particular, see the included LICENSE.md file.

Bugs and questions

Bug reports and feedback are always welcome. For reporting bugs and requesting minor improvements, please use the "Issues" functionality in Github. For larger requests for improvement or jolly scholarly co-operation in expanding the true score imputation framework, please email me directly at [e-mail]. When in doubt, either contact method is fine.

mmansolf/TSI documentation built on Aug. 29, 2023, 2:38 a.m.