The purpose of matching is to approximate an experimental template, where the matching procedure approximates blocking prior to random treatment assignment in order to balance covariates between treatment and control groups. Separation of the estimation procedure into two steps simulates the research design of an experiment, where no information on outcomes is known at the point of experimental design and randomization. The separation of the balancing process in MATCHIT from the analysis process afterward helps keep clear the goal of balancing control and treatment groups and makes it less likely that the user will inadvertently cook the books in his or her favor.