Imputation Software

The Blimp application for Mac and Windows performs single-level and multilevel imputation with mixtures of continuous, ordinal, and nominal variables.  Blimp implements a fully conditional specification algorithm (also known as chained equations and sequential regression) with a latent variable formulation for incomplete categorical variables.  Blimp was developed with funding from Institute of Educational Sciences award R305D150056 (Craig Enders, PI).  Algorithmic development by Craig Enders and Brian Keller, C++ programming by Brian Keller, and graphical user interface development by Nitish Mehta and Brian Keller.

NEW!  Blimp 1.1 is now available with exciting new features, including an imputation routine (substantive model compatible imputation) for accurate imputation of interactive and polynomial effects, and improved estimation of random slopes in multilevel models with up to three levels.

Blimp Application Download

Blimp Documentation and Support

User Guide Examples

The Blimp User Guide illustrates imputation for several common single-level and multilevel analyses. Each example comes with a zip archive that includes slides explaining the analysis, the raw data files, Blimp scripts, and analysis scripts for Mplus, R, SAS, SPSS, and Stata. 

  • Example 1: Single-level regression analysis
  • Example 2Single-level regression with interaction effect
  • Example 3Two-level regression with random intercepts
  • Example 4: Two-level regression with random slopes
  • Example 5Two-level regression with random slopes and a cross-level interaction effect
  • Example 6: Three-level regression with random intercepts
  • Example 7: Three-level regression with random slopes and cross-level interaction effect
  • Example 8.2: R code for an external simulation with Blimp

Blimp Papers and Materials

Enders, C.K., Du, H. Keller, B.T. (2018). Substantive model-compatible imputation for multilevel regression models with random coefficients, interaction effects, and other nonlinear terms. Paper submitted for publication.

  • Click here to request a copy of the paper

Enders, C.K. (2018, February). A model-based imputation procedure for moderated regression analyses. Colloquium presented to the Quantitative Forum, Vanderbilt University, Nashville, TV.

  • Click here to download the presentation

Enders, C., Du, H., & Keller, B. (2017, October). A fully Bayesian imputation procedure for random coefficient models (and other pesky product terms). Paper presented at the Society of Multivariate Experimental Psychology, Minneapolis, MN.

  • Click here to download the presentation

Enders, C.K., Keller, B.T., & Levy, R. (2017). A fully conditional specification approach to multilevel imputation of categorical and continuous variables. Psychological Methods. Advance online publication. http://dx.doi.org/10.1037/met0000148.

  • Click here to download the paper
  • Click here to download the technical appendix
  • Click here to download the data file and analysis scripts

Enders, C.K. (2016). Multiple imputation as a flexible tool for missing data handling in clinical research. Behaviour Research and Therapy. Advanced online publication. http://dx.doi.org/10.1016/j.brat.2016.11.008.

  • Click here to download the paper
  • Click here to download the raw data file, imputed data sets, Blimp script, and R analysis script

Version History

Blimp 1.1 (4.2.18)

  • Additional error reporting and printing improvements
  • Minor bug fixes and algorithmic improvements
  • Fixed a bug where certain configurations of interactions were not being recognized in syntax
  • Fixed a bug that sometimes caused nominal variables to be omitted from output reporting
  • Tabs will now be treated as spaces in the syntax parsing
  • PRIOR2 is now the default for SMC-FCS (i.e. OUTCOME command)
  • Error message added to flag a non-numeric missing value code

Blimp 1.0 (9.13.17)

  • Substantive model-compatible fully conditional specification (SMC-FCS)
  • Interactive and polynomial effects via SMC-FCS
  • BYGROUP command for separate-group imputation
  • New User’s Guide including worked examples with analysis scripts for Mplus, R, SAS, SPSS, and Stata
  • Expanded algorithmic options
  • Syntax convenience features
  • Minor bug fixes and algorithmic improvements
Questions or suggestions? Email Craig Enders