What’s new in version 1.3-1 (built on 2012-12-07) *Added data directory with .rda files for wcgs and schisto data can be loaded with data(schisto) or data(wcgs) *Parallel processing of bootstrap replicates now uses parallel package (distributed as part of core R as of R version 2.14.0) i.e. packages multicore and rlecuyer are no longer required when multiPIMboot is called with multicore = TRUE. *argument rlecuyer.seed to multiPIMboot function has been replaced with the argument mc.seed, which should be a single integer *Added timing information to multiPIM results *Added type argument to summary.multiPIM and summaries now by default include breakdown of running time by g vs. Q modeling and by super learner candidate in addition to the statistical summary by exposure-outcome pair *Added this ChangeLog file *Split up file R/functions.R into three separate files by functionality: -multiPIM.R: definitions of the candidate vectors, multiPIM function -multiPIMboot.R: multiPIMboot function -summary.R: summary.multiPIM and print.summary.multiPIM functions What’s new in version 1.2-1 (2011-11-02): *Minor updates for compatibility with R version 2.14 *mutliPIMboot no longer allows for automatic detection of number of cores to use when multicore = FALSE What’s new in version 1.1-2 (2011-09-20): *Bug fix: there was an error in the formula which calculates plug-in standard errors for IPCW and DR-IPCW estimators. This has now been fixed. What’s new in version 1.1-1 (2011-09-14): *New estimators: the default estimator is now TMLE. Also, a G-COMP estimator is now available, as well as the IPCW and DR-IPCW estimators that were available in the previous version. *Default for g.method has been changed from super learning to main terms logistic regression. This will usually work better with the new TMLE estimator. *Support for factors in W. *standardize argument for use with the lars and penalized candidates.