title:         Alternative Balance Metrics for Bias Reduction in Matching Methods for Causal Inference
authors:       Jasjeet Sekhon
entrydate:     2006-07-18 21:20:00
keywords:      Matching, Causal Inference, Genetic Matching, Balance Metrics
abstract:      Sekhon (2006; 2005) and Diamond and Sekhon (2005)
propose a matching method, called Genetic
Matching, which algorithmically maximizes the
balance of covariates between treatment and
control observations via a genetic search
algorithm (Sekhon and Mebane 1998). The method is
neutral as to what measures of balance one wishes
to optimize. By default, however, a variety of
hypothesis tests are used as balance metrics
although it is made clear that these are not
formal tests. Imai, King, and Stuart (2006) argue
that instead of using hypothesis tests (such as
t-tests) to evaluate balance, researchers should
use descriptive measures of discrepancy (such as
functions of empirical QQ plots) which they claim
are monotonic functions of balance. However, such
descriptive measures are not monotonic functions
of bias in the estimand of interest and they
ignore key information related to bias. Using
descriptive metrics, one is unable to reliably
recover the experimental benchmark in a testbed
dataset for matching estimators (Dehejia and Wahba
1999). And these metrics, unlike those based on
non-parametric hypothesis tests, perform poorly in
a series of Monte Carlos just as one would expect
given their properties.

http://polmeth.wustl.edu/retrieve.php?id=625

**********************************************************
             Political Methodology E-Mail List
        Editor: Karen Long Jusko <[log in to unmask]>
**********************************************************
        Send messages to [log in to unmask]
  To join the list, cancel your subscription, or modify
           your subscription settings visit:

          http://polmeth.wustl.edu/polmeth.php

**********************************************************