Statistical matching: hot deck or propensity score?

This empirical study compares hot deck (HD) and propensity score (PS) statistical matching methods (SM) with the aim of providing useful guidance to researchers interested in implementing SM methods. Our main goal is to narrow the methodological gap between the two approaches to statistical matching. In our empirical setting, HD and PS yield similar results. We argue that the choice between HD or PS is not critical from a statistical point of view if the method is properly executed. The comparison of the HD and PS methods in this study shows the importance of unifying the two methods while minimizing the execution uncertainties caused by non-optimal modeling decisions made by practitioners.