stat_match

Statistical matching is a statistical procedure by which a sub-sample of treated units, i.e. that have received some treatment, is matched with a sub-sample of untreated units which should represent the first if they had not received the treatment. This procedure is one of the methods to produce an estimate of the causal effect of the treatment. Among the main statistical-econometric software, Gretl was the only one that did not provide a package to perform statistical matching. With the package "stat_match" this gap has been filled. The actual version of the package allows to perform two matching methods: (i) radius matching: each treated unit is matched with untreated units whose proximity measure is less than a predetermined limit (caliper); (ii) one-to-one matching: each treated unit is matched with the nearest untreated unit, in terms of proximity measure. These methods are available with both propensity score and mahalanobis distance.