Bootstrapping for propensity score analysis and matching.
Fixed new R CMD CHECK notes and warnings.
Updated for new version of party package.
Refactored the function for calculating balance to be shared across all the matching procedures.
Added a vignette about the impact of matching order.
Fixed an issue where under certain circumnstances there could be a NA effect estimate. These are removed.
Added better error checking for when a classification tree with no splits occurs.
The summary function now includes a weighted pooled estimate. This estimate is weighted using the inverse balance estimates so that boostrap samples with better balance are weighted more to the estimate than those with worse balance.
You can now specify how the balance estimates are aggregated. Possible values include mean (the default), q25, q75, median, or max.
Added getPSAbootMethods and global R option "PSAboot.methods" that will return the default set of functions to use in PSAboot.
Added a number of parameters to the boxplot function including a tufte style from the ggthemes package.
Initial version of PSAboot package for bootstrapping propensity score analysis.