TY - JOUR
AU1 - Schafer, Joseph L.
AB - This chapter offers an accessible introduction to missing-data procedures, especially for longitudinal data, using multiple imputation (Rubin, 1987; Shafer, 1997a), which is the practice of replacing missing data with plausible values. The author points out that even if imputation methods successfully preserve important aspects of the data distributions, a potentially serious problem remains in that imputation adds fictitious information to a data set. If imputed values are treated the same way as observed values in subsequent analyses, then the resulting inferences will be artificially precise. The key idea of multiple imputation is that it treats missing data as an explicit source of random variability over which to be averaged. The process of creating imputations, analyzing the imputed data sets, and combining the results is a Monte Carlo version of averaging the statistical results over the predictive distribution. The author describes a method for creating multiple imputations in longitudinal databases, using PAN, a library of algorithms developed for imputing multivariate panel data. Comments by A. Davey on this chapter and that of Graham et al (see record 2001-01077-011) follows Chapter 12 on pp. 379–383. (PsycInfo Database Record (c) 2024 APA, all rights reserved)
TI - New methods for the analysis of change.: Multiple imputation with PAN.
DA - 2004-08-31
UR - https://www.deepdyve.com/lp/american-psychological-association/new-methods-for-the-analysis-of-change-multiple-imputation-with-pan-8iaHD02h9I
DP - DeepDyve
ER -