Multivariate matching methods that are equal percent bias reducing

By Donald B. Rubin. 1974

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

A Potential Outcomes View of Value-Added Assessment in Education (2004, methodology): “…if the missing data are missing completely at random (MCAR, Little & Rubin, 2002) and is not appropriate in settings such as this with longitudinal data and missing outcomes (see, e.g., Barnard et al., 1998, 2003), even when the missing data are ignorable (Rubin, 1976c; Little and Rubin, 2002).”
Rerandomization to improve covariate balance in experiments (2012, methodology): “Analogous to “Equal Percent Bias Reducing” (EPBR) matching methods [Rubin (1976)], a rerandomization method is said to be “Equal Percent Variance Reducing” (EPVR) if the percent reduction in variance is the same for each covariate.”
Using Propensity Scores to Help Design Observational Studies: Application to the Tobacco Litigation (2001, methodology): “A key bridge between matching and propensity scores is the use of a one-dimensional summary for matching, specifically ‘‘discriminant matching’’ (Cochran and Rubin, 1973; Rubin, 1976a,b, 1979, 1980).”
For objective causal inference, design trumps analysis (2008, background): “Here is where traditional power calculations are relevant; also extensions, for example, involving the ratios of sample sizes needed to obtain well-matched samples [Rubin (1976b), Section 5], are relevant, and should be considered before plunging ahead.”
Generalized Optimal Matching Methods for Causal Inference (2016, methodology): “4.6 and on connections to equal percent bias reduction (EPBR) in Appendix A. RKHS norms always satisfy the conditions of Def.”
Affinely invariant matching methods with discriminant mixtures of proportional ellipsoidally symmetric distributions (2006, methodology): “(a) When E(Zrt − Zrc) 6= 0, the matching is equal percent bias reducing (EPBR), as defined by [14],”
Matching methods for causal inference: A review and a look forward (2010, methodology): “To formalize, using notation similar to that in Rubin (1976b), we consider two populations, Pt and Pc, where the subscript t refers to a group exposed to the treatment and c refers to a group exposed to the control.”
Optimal Full Matching and Related Designs via Network Flows (2006, background): “ment may not be appropriate, a theme with echoes in later work of Rubin (1976) and his”
MALTS: Matching After Learning to Stretch (2018, result): “in congruence with consistency results by Stone (1977) and Jiang (2019); those works handled the special case where the weights are uniform probability weights instead of any probability weights.”
Matching With Multiple Control Groups With Adjustment for Group Differences (2008, unknown): “…−XmcÞ+ dEðDt −DmcÞ = bðmt −EðXmcÞÞ+ dð0− Nt −mNt Þ = bmt − b mNt mc1 + m Nt p 4 sc1log Nc1 m + Nt −m Nt mt + Nt −mNt p 4 stlog Nt Nt −m + d m−Nt Nt : This formula uses the approximation for the tail expectation of a univariate standard normal distribution from Rubin (1976), (N, n)≈ p 4 lnðN n Þ.”