Variable-Ratio Matching with Fine Balance in a Study of Peer Health Exchange

Luke Keele,Sam Pimentel,Frank Yoon

DOI: https://doi.org/10.48550/arXiv.1410.4723

2014-10-17

Methodology

Abstract:In observational studies of treatment effects, matched samples are created so treated and control groups are similar in terms of observable covariates. Traditionally such matched samples consist of matched pairs. If a pair match fails to make treated and control units sufficiently comparable, alternative forms of matching may be necessary. One general strategy to improve balance is to match a variable number of control units to each treated unit. A more tailored strategy is to adopt a fine balance constraint. Under a fine balance constraint, a nominal covariate is exactly balanced, but it does not require individually matched treated and control subjects for this variable. In the example, we seek to construct a matched sample for an ongoing evaluation of Peer Health Exchange, an intervention in schools designed to decrease risky health behaviors among youth. We find that an optimal pair match that minimizes distances between pairs creates a matched sample where balance is poor. Here we propose a method to allow for fine balance constraints when each treated unit is matched to a variable number of control units, which is not currently possible using existing matching algorithms. Our approach uses the entire number to first determine the optimal number of controls for each treated unit. For each strata of matched treated units, we can then apply a fine balance constraint. We then demonstrate that a matched sample for the evaluation of the Peer Health Exchange based on a variable number of controls and fine balance constraint is superior to simply using a variable-ratio match.