NEWS.md
In MatchIt.mice: Matching Multiply Imputed Datasets

MatchIt.mice

Matching Multiply Imputed Datasets

The MatchIt.mice package gets a massive update! Now, you can match treatment and control observations within each imputed dataset, resulting in several estimates of treatment effect size (obtained from the complete case analysis of each imputed dataset), and then pool these several treatment effect size estimates. This approach is called within (or match-then-pool). Alternatively, you can average each observation's distance measures (propensity scores) from different imputed datasets, match treatment and control observations based on their averaged distance measures, and then average the treatment effect estimates obtained from the complete case analysis of each imputed dataset. This approach is called across (or pool-then-match). Minor bugs are fixed and the README.md and NEWS.md files are updated.

The mergeitmice() function is now included in the package, which merges mids, mimids, and wimids objects with a dataframe. The binditmice() function is also updated to work with mimids and wimids objects. The pool() function is updated to produce more reliable results when numbers of observations in matched datasets are different. The matchitmice() function is also updated to sort results before returning the output (as thus, matchitmice.data() function is also updated). Minor bugs are fixed.

Minor bugs are fixed.

The matchitmice() and the weightitmice() functions are updated to match and weight imputed datasets, their outputs will be saved in the mimids and wimids class objects, and the plot(), print(), and summary() functions are updated to be able to provide detailed descriptions of these objects. The matchitmice.data() and weightitmice.data() functions are added for extracting matched and weighted imputed datasets from the mimids and wimids class objects and to replace the retired matchmicedata() function. The with() function now works with the mimids and wimids class objects and the pool() function can be used to pool the obtained results from analyses on each imputed dataset. Examples are simplified, the reference manual is updated, the README.md and NEWS.md files are revised, and plenty of minor bugs are fixed.