The DOI matches its Crossref record.
Comparison of Multivariate Matching Methods: Structures, Distances, and Algorithms
Abstract
A comparison and evaluation is made of recent proposals for multivariate matched sampling in observational studies, where the following three questions are answered: (1) Algorithms: In current statistical practice, matched samples are formed using “nearest available” matching, a greedy algorithm. Greedy matching does not minimize the total distance within matched pairs, though good algorithms exist for optimal matching that do minimize the total distance. How much better is optimal matching than greedy matching? We find that optimal matching is sometimes noticeably better than greedy matching in the sense of producing closely matched pairs, sometimes only marginally better, but it is no better than greedy matching in the sense of producing balanced matched samples. (2) Structures: In common practice, treated units are matched to one control, called pair matching or 1–1 matching, or treated units are matched to two controls, called 1–2 matching, and so on. It is known, however, that the optimal structure is a full matching in which a treated unit may have one or more controls or a control may have one or more treated units. Optimal 1 — k matching is compared to optimal full matching, finding that optimal full matching is often much better. (3) Distances: Matching involves defining a distance between covariate vectors, and several such distances exist. Three recent proposals are compared. Practical advice is summarized in a final section.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
No references could be resolved yet.
Research works printed but not linked: 19
- Anderson , T. 1984 . An Introduction to Multivariate Statistical Analysis, , second ed. , New York : John Wiley .
- Bertsekas , D. 1991 . Linear Network Optimization: Algorithms and Codes , Cambridge , MA : MIT Press .
- 10.1093/biomet/64.2.299 · DOI
- Cochran , W. G. 1965 . “The Planning of Observational Studies of Human Populations” . Journal of the Royal Statistical Society , 128 : 134 – 155 . (with discussion), Ser. A
- Cochran , W. G. and Rubin , D. B. 1973 . “Controlling Bias in Observational Studies: A Review” . Sankya , 35 : 417 – 446 . Ser. A
- 10.1016/0002-9149(81)90391-X · DOI
- Gu , X. 1992 . “A Comparison and Evaluation of Recent Developments for Multivariate Matching in Observational Studies” , University of Pennsylvania, Department of Statistics . unpublished Ph.D. thesis
- 10.2307/1165073 · DOI
- 10.1093/biomet/75.3.577 · DOI
- 10.1080/01621459.1989.10478868 · DOI
- 10.1111/j.2517-6161.1991.tb01848.x · DOI
- 10.1093/biomet/70.1.41 · DOI · Paper page
- 10.1080/00031305.1985.10479383 · DOI
- 10.2307/2529685 · DOI
- Rubin , D. B. 1976 . “Multivariate Matching Methods That are Equal Percent Bias Reducing, I: Some Examples” . Biometrics , 32 : 185 – 203 .
- 10.1080/01621459.1979.10482513 · DOI
- 10.2307/2529981 · DOI
- 10.2307/2529548 · DOI · Paper page
- 10.1137/0208036 · DOI
Cited by
Sorted by citations
Items 201–300 of 595
Civic Talk and Civic Participation
Klofstad, C. (2009) · American Politics Research · doi:10.1177/1532673x09333960 · Cited by 53
“…and characteristics. at DUKE UNIV on August 17, 2009 http://apr.sagepub.comDownloaded from For this analysis a “full matching” procedure was used (Gu & Rosenbaum, 1993; Hansen 2004; Ho et al., 2007a, 2007b; Rosenbaum, 1991).6 Full matching is a hybrid of “subclassification” and “optimal…”
MethodThe Long-Run Effects of R&D Place-Based Policies: Evidence from Russian Science CitiesPreprint
Schweiger, H., Stepanov, A., & Zacchia, P. B. (2022) · American Economic Journal: Economic Policy · doi:10.2139/ssrn.3203317 · Cited by 32
Is incineration repressing recycling?
Yamamoto, M., & Kinnaman, T. C. (2021) · Journal of Environmental Economics and Management · doi:10.1016/j.jeem.2021.102593 · Cited by 37
Personalized pricing for customer retention: Theory and evidence from mobile communication
Capponi, G., Corrocher, N., & Zirulia, L. (2020) · Telecommunications Policy · doi:10.1016/j.telpol.2020.102069 · Cited by 45
Hannan, E., Zhong, Y., Berger, P., Walford, G., Curtis, J., Wu, C., Venditti, F. J., Higgins, R., Smith, C. R., Lahey, S., et al. (2014) · American Journal of Cardiology · doi:10.1016/j.amjcard.2013.11.035 · Cited by 32
Using Propensity Score Analysis for Making Causal Claims in Research Articles
Bai, H. (2011) · Educational Psychology Review · doi:10.1007/s10648-011-9164-9 · Cited by 28
“In terms of the effectiveness of PSA, Gu and Rosenbaum (1993) found that propensity score matching usually outperformed other matching techniques because it balanced on many covariates simultaneously, which potentially approximated the balance achieved through randomization.”
BackgroundMethodInfluentialConditional Causal Relationships between Emotions and Causes in Texts
Chen, X., Li, Q., & Wang, J. (2020) · Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2020.emnlp-main.252 · Cited by 31
“…discovered and “removed” by some specially designed algorithms in order to accurately estimate the causal effect, such as propensity score method (Gu and Rosenbaum, 1993; Austin, 2011; Imbens, 2004; Lunceford and Davidian, 2004), front-door criterion (Pearl, 1995), instrumental variable…”
MethodHannan, E., Samadashvili, Z., Smith, C. R., Lahey, S., Gold, J., Jordan, D., Sundt, T., Girardi, L., Ashraf, M. H., & Chikwe, J. (2019) · Journal of Thoracic and Cardiovascular Surgery · doi:10.1016/j.jtcvs.2018.08.091 · Cited by 31
“The propensity score was used to match patients on a 1-to-1 basis so as to minimize the overall distance in propensity scores between the groups.(16-19) Patient characteristics for MVR and MV-repair patients were matched as long as their estimated log-odds from the logistic regression model were no more than 0.”
MethodMALTS: Matching After Learning to StretchPreprint
Parikh, H., Rudin, C., & Volfovsky, A. (2018) · Journal of machine learning research · doi:10.48550/arxiv.1811.07415 · Cited by 31
“Other causal inference methods have been studied in the literature (Gu and Rosenbaum 1993, Imbens 2004), but almost all of them suffer from at least one of four possible problems: using a black box model that is uninterpretable (i.e., almost all machine learning methods), having a distance metric…”
BackgroundCafri, G., Wang, W., Chan, P. H., & Austin, P. C. (2018) · Statistical Methods in Medical Research · doi:10.1177/0962280218799540 · Cited by 29
Propensity score methodology for confounding control in health care utilization databases
Patorno, E., Grotta, A., Bellocco, R., & Schneeweiß, S. (2022) · Epidemiology Biostatistics and Public Health · doi:10.2427/8940 · Cited by 57
“PROPENSITy SCORE fOR CONfOuNDING ADjuSTMENT IN HEALTH CARE DATABASES one treated subject and at least one untreated subject or one untreated subject and at least one treated subject, has also been proposed [52-54].”
BackgroundPinto, D. S., Ogbonnaya, A., Sherman, S., Tung, P., & Normand, S. (2012) · Circulation. Cardiovascular Quality and Outcomes · doi:10.1161/circoutcomes.111.961938 · Cited by 31
The Lasting Effect of Civic Talk on Civic Participation: Evidence from a Panel Study
Klofstad, C. (2010) · Social Forces · doi:10.1353/sof.2010.0047 · Cited by 51
“In this paper, a “full matching” data preprocessing procedure was used (Gu and Rosenbaum 1993; Hansen 2004; Ho et al. 2007a; Rosenbaum 1991).”
MethodSánchez‐López, M., Cavero‐Redondo, I., Álvarez‐Bueno, C., Hermosa, A. R., Pozuelo‐Carrascosa, D. P., Díez‐Fernández, A., Gutiérrez, D., Pardo‐Guijarro, M. J., & Martínez‐Vizcaíno, V. (2019) · Scandinavian Journal of Medicine and Science in Sports · doi:10.1111/sms.13383 · Cited by 40
Oliveira, G. D. D., Bialek, J. M., Nicosia, L., McCarthy, R., Chang, R., Fitzgerald, P., & Kim, J. Y. S. (2014) · Breast · doi:10.1016/j.breast.2014.01.003 · Cited by 30
“6 SD will remove approximate 90% of the bias in observed confounders [21].”
BackgroundGriffin, J. R., Lee, S., Caiado, M., Kesten, S., & Price, D. (2008) · Primary Care Respiratory Journal · doi:10.3132/pcrj.2008.00024 · Cited by 30
Fu, A. Z., Dow, W. H., & Liu, G. G. (2007) · Health Services and Outcomes Research Methodology · doi:10.1007/s10742-006-0016-x · Cited by 25
“A simulation suggests that matching on the propensity score is best when there are more than 20 covariates (Gu and Rosenbaum 1993).”
Gatekeeping Under Congestion: An Empirical Study of Referral Errors in the Emergency Department
Freeman, M., Robinson, S. M., & Scholtes, S. (2017) · Management Science · doi:10.1287/mnsc.2020.3711 · Cited by 45
“This condition is a conservative distance that has been shown to reduce more than 90% of bias due to observable differences between treatment and control groups (Gu and Rosenbaum 1993).”
BackgroundCharytan, D., Varma, M. R., Silbaugh, T. S., Lovett, A. F., Normand, S., & Mauri, L. (2011) · American Journal of Kidney Diseases · doi:10.1053/j.ajkd.2010.09.017 · Cited by 29
“Standardized differences 10% supported the assumption of balance between treatment groups.(23)”
BackgroundMethodThe association between adolescent football participation and early adulthood depression
Deshpande, S. K., Hasegawa, R. B., Weiss, J., & Small, D. S. (2020) · PLoS ONE · doi:10.1371/journal.pone.0229978 · Cited by 32
“To achieve a good compromise between overall covariate balance and similarity of matched subjects, we matched using a propensity score-calipered rank-based Mahalanobis distance between the baseline covariates of each pair of exposed and control subject [49].”
MethodKim, S., Cheon, S., & Suh, H. S. (2019) · The Annals of Pharmacotherapy · doi:10.1177/1060028019859543 · Cited by 28
Risk of Suicidal Behavior With Antidepressants in Bipolar and Unipolar Disorders
Leon, A. C., Fiedorowicz, J. G., Solomon, D. A., Li, C., Coryell, W. H., Endicott, J., Fawcett, J., & Keller, M. B. (2014) · The Journal of Clinical Psychiatry · doi:10.4088/jcp.13m08744 · Cited by 33
Cattran, D., Reich, H., Kim, S., & Troyanov, S. (2011) · American Society of Nephrology. Clinical Journal · doi:10.2215/cjn.11001210 · Cited by 28
“05 difference, a conservative value (24).”
On the use of discrete choice models for causal inference
Tchernis, R., Horvitz-Lennon, M., & Normand, S.-L. T. (2005) · Statistics in Medicine · doi:10.1002/sim.2095 · Cited by 28
“Under these assumptions, several methods can be used to adjust for X, including matched sampling [17].”
MethodNúñez, E., Arnett, D., Benjamin, E. J., Oakes, J., Liebson, P., & Skelton, T. (2004) · American Journal of Cardiology · doi:10.1016/j.amjcard.2004.08.012 · Cited by 28
Leckelt, M., Richter, D., Schröder, C., Küfner, A. C. P., Grabka, M. M., & Back, M. D. (2019) · British Journal of Psychology · doi:10.1111/bjop.12360 · Cited by 27
“To this end, we used theMatchIt package version 2.4-21 for R (Ho, Imai, King, & Stuart, 2011) and employed a full-matching procedure (Gu & Rosenbaum, 1993) that had the advantage of being able to use all cases in the samples for the matching.”
MethodInfluentialChen, K., Kuo, S., Chien, P., Hsieh, H.-Y., & Hsieh, C.-H. (2021) · Scientific Reports · doi:10.1038/s41598-021-82034-5 · Cited by 27
Lee, S. J., Jeung, K., Lee, B., Min, Y., Park, K., Suh, G., Kim, K. S., & Kang, G. (2015) · American Journal of Emergency Medicine · doi:10.1016/j.ajem.2014.10.003 · Cited by 27
School Choice in NY City: A Bayesian Analysis of an Imperfect Randomized Experiment
Barnard, J., Frangakis, C., Hill, J. L., & Rubin, D. (2001) · Lecture notes in statistics · doi:10.1007/978-1-4613-0035-9_1 · Cited by 27
“Mahalanobis matching is most effective when applied to a small number of essentially continuous covariates (Rosenbaum and Rubin 1985; Gu and Rosenbaum 1993).”
BackgroundEsper, S., Holder-Murray, J., Subramaniam, K., Boisen, M., Kenkre, T., Meister, K., Foos, S., Wong, H. P., Howard-Quijano, K., & Mahajan, A. (2020) · Annals of Surgery · doi:10.1097/sla.0000000000004642 · Cited by 26
A Framework for Optimal Matching for Causal Inference
Kallus, N. (2016) · International Conference on Artificial Intelligence and Statistics · doi:10.48550/arxiv.1606.05188 · Cited by 26
“one composesamatchedsampleoutofpairsoftreatedandcontrol subjects so that the total pairwise distance between covariate vectors is small or even minimal, mimicking a randomized matched-pair experiment [14]. If we allow subjects to be paired with replacement, we can have a sample with duplicates. Weighting is a generali…”
BackgroundHigh levels of uPA and PAI-1 predict a good response to anthracyclines
Borštnar, S., Sadikov, A., Možina, B., & Čufer, T. (2010) · Breast Cancer Research and Treatment · doi:10.1007/s10549-009-0691-8 · Cited by 26
“The matching algorithms in the literature, surveyed by Gu and Rosenbaum [25], usually pick tuples of patients in each step—one (or more) belonging to each group that differ less than some prescribed threshold between one another.”
Method6 Observational studies and nonrandomized experiments
Rosenbaum, P. R. (1996) · Handbook of statistics · doi:10.1016/s0169-7161(96)13008-x · Cited by 28
Self-citationPuck, J. F., Hödl, M. K., Filatotchev, I. V., Wolff, H., & Bader, B. (2016) · Asia Pacific Journal of Management · doi:10.1007/s10490-016-9471-2 · Cited by 25
“Although our sample appears to be small in comparison to large scale surveys that have employed this method, others have shown that matching reduces bias substantially even in small samples (e.g., Gu & Rosenbaum, 1992 use 50 units in each, treatment and control group, see also Rubin, 1979).”
ResultLorenzo, G. D., Casciano, R., Malangone, E., Buonerba, C., Sherman, S., Willet, J., Wang, X., Liu, Z.-M., & Placido, S. D. (2011) · Expert Opinion on Pharmacotherapy · doi:10.1517/14656566.2011.587119 · Cited by 25
“As no head-to-head drug comparison of everolimus versus sorafenib after failure with sunitinib is currently available, indirect comparisons may provide valuable information regarding comparative effectiveness with the drugs [10].”
BackgroundDistance functions for matching in small samples
Dettmann, E., Becker, C., & Schmeißer, C. (2010) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2010.11.022 · Cited by 22
“Thematching process, but not the estimation result itself, is examined bymeans of the percent bias reduction (Augurzky, 2000b; Cochran, 1968; Cochran and Rubin, 1973; Gu and Rosenbaum, 1993).”
BackgroundMethodInfluentialSchool Improvement Plans and Student Learning in Jamaica
Lockheed, M. E., Harris, A., & Jayasundera, T. (2010) · doi:10.1016/j.ijedudev.2009.06.007 · Cited by 25
Funch, D., Mortimer, K., Ziyadeh, N. J., Seeger, J. D., Zhou, L., Ng, E., Ross, D., Major-Pedersen, A., Bosch-Traberg, H., Gydesen, H., et al. (2021) · Diabetes, Metabolic Syndrome and Obesity: Targets and Therapy · doi:10.2147/dmso.s305496 · Cited by 24
Causal inference in randomized clinical trials
Zheng, C., Dai, R., Gale, R., & Zhang, M.-J. (2019) · Bone Marrow Transplantation · doi:10.1038/s41409-018-0424-x · Cited by 24
Leite, W. L., Stapleton, L. M., & Bettini, E. A. (2018) · Structural Equation Modeling A Multidisciplinary Journal · doi:10.1080/10705511.2018.1522591 · Cited by 28
“Studies comparing matching strategies (e.g. Austin, 2009, 2014; Cepeda, Boston, Farrar, & Strom, 2003; Diamond & Sekhon, 2013; Gu & Rosenbaum, 1993) have shown that optimal, genetic algorithms or the use of a caliper produce the best covariate balance and reduction in bias in most conditions.”
Weingarten, T., Taccolini, A. M., Ahle, S. T., Dietz, K., Dowd, S. S., Frank, I., Boorjian, S., Thapa, P., Hanson, A., Schroeder, D., et al. (2016) · Canadian Journal of Anesthesia/Journal canadien d'anesthésie · doi:10.1007/s12630-016-0599-9 · Cited by 24
Connealy, N. T. (2020) · Crime and delinquency · doi:10.1177/0011128720926116 · Cited by 23
“…to its more common application of balancing covariates and independent variables (Schnell et al., 2019).10 The present study’s application of PSM invoked nearest neighbor matching, which has proved to be a valid PSM technique for generating one-to-one matches (Austin, 2011; Gu & Rosenbaum, 1993).”
Wang, C., Li, K.-G., Kim, M., Lee, S., & Seo, D. (2019) · Addictive Behaviours · doi:10.1016/j.addbeh.2018.10.037 · Cited by 23
Mercury pollution, information, and property values
Tang, C., Heintzelman, M., & Holsen, T. (2018) · Journal of Environmental Economics and Management · doi:10.1016/j.jeem.2018.10.009 · Cited by 23
Matching algorithms for causal inference with multiple treatments
Scotina, A. D., & Gutman, R. (2019) · Statistics in Medicine · doi:10.1002/sim.8147 · Cited by 20
“These results are similar to the results for matching on the Mahalanobis distance of the covariates for binary treatments, where the performance of matching on the Mahalanobis distance deteriorate with increasing P [33, 34].”
BackgroundResultLee, J., Park, K.-W., Han, J., Yang, H.-M., Kang, H.-J., Koo, B., Bae, J., Woo, S., Park, J., Jin, D., et al. (2014) · American Journal of Cardiology · doi:10.1016/j.amjcard.2014.07.065 · Cited by 23
Cepeda, M., Boston, R., Farrar, J., & Strom, B. (2003) · Journal of Clinical Epidemiology · doi:10.1016/s0895-4356(02)00583-8 · Cited by 23
“The larger the percent reduction in bias, the better the method is [7,8].”
BackgroundMethodInfluentialAnalysis Approaches To Community Evaluation
Gruenewald, P. (1997) · Evaluation review · doi:10.1177/0193841x9702100205 · Cited by 23
Fiscal effects of budget referendums: evidence from New York school districts
Nguyen-Hoang, P. (2012) · doi:10.2139/ssrn.1971559 · Cited by 23
“Put differently, propensity score matching is intended to address what Heckman and Robb (1985) called “selection on observables,” i.”
BackgroundMethodInfluentialCardiac Troponin and Recurrent Major Vascular Events after Minor Stroke or Transient Ischemic Attack
Hellwig, S., Ihl, T., Ganeshan, R., Laumeier, I., Ahmadi, M., Steinicke, M., Weber, J., Endres, M., Audebert, H., & Scheitz, J. (2021) · Annals of Neurology · doi:10.1002/ana.26225 · Cited by 22
“To further account for selection bias, we performed propensity score matching (PSM) in a 1:1 fashion without replacement (“optimal matching”) to balance baseline characteristics between patients with hs-cTnT below or above URL.23,24 We used a caliper (ie, the maximum distance that two cases can be apart from each othe…”
MethodInfluentialFulfillment by Amazon versus fulfillment by seller: An interpretable risk‐adjusted fulfillment model
Sun, L., Lyu, G., Yu, Y., & Teo, C. (2020) · Naval Research Logistics (NRL) · doi:10.1002/nav.21897 · Cited by 31
“Following a wide stream of observational studies (Abadie & Imbens, 2016; Gu & Rosenbaum, 1993; Hansen, 2004; Ming & Rosenbaum, 2000), we incorporate PSM algorithm in our study to address the data selection bias issue existing in the FBA vs FBS fulfillment channel selection problem, considering that…”
MethodBank Loan Supply Responses to Federal Reserve Emergency Liquidity FacilitiesPreprint
Berger, A. N., Black, L. K., Bouwman, C. H. S., & Dlugosz, J. L. (2017) · SSRN Electronic Journal · doi:10.2139/ssrn.2429710 · Cited by 22
“This is a one-sided version of what is generally known as a greedy matching algorithm (Gu and Rosenbaum, 1993).”
Hannan, E., Samadashvili, Z., Stamato, N., Lahey, S., Wechsler, A., Jordan, D., Sundt, T., Gold, J., Ruiz, C., Ashraf, M. H., et al. (2016) · JACC: Cardiovascular Interventions · doi:10.1016/j.jcin.2015.12.022 · Cited by 22
Fischer, J., Wes, A., Kanchwala, S. K., & Kovach, S. (2014) · Journal of Plastic Surgery and Hand Surgery · doi:10.3109/2000656x.2013.877915 · Cited by 22
Wu, C., Dyer, A., Walford, G., Holmes, D., King, S., Stamato, N., Sharma, S. K., Jacobs, A., Venditti, F. J., & Hannan, E. (2013) · American Journal of Cardiology · doi:10.1016/j.amjcard.2013.05.003 · Cited by 22
Vo, Y., Rickels, H., Welch, C., & Dunbar, S. (2023) · Assessing Writing: An International Journal · doi:10.1016/j.asw.2023.100719 · Cited by 21
Randomization Tests to Assess Covariate Balance When Designing and Analyzing Matched Datasets
Branson, Z. (2021) · Observational Studies · doi:10.1353/obs.2021.0031 · Cited by 22
“…focus on the standardized covariate mean differences (Stuart, 2010; Zubizarreta, 2012) and Mahalanobis distance (Mahalanobis, 1936; Rosen- baum & Rubin, 1985; Gu & Rosenbaum, 1993; Diamond & Sekhon, 2013), because they are the most commonly used measures for appraising balance in matched datasets.”
MethodZhong, H.-T., Li, W., & Boarnet, M. (2019) · Environment and Planning B Urban Analytics and City Science · doi:10.1177/2399808320982305 · Cited by 21
“We start with a data generating process (DGP) that is similar to Gu and Rosenbaum (1993) and King and Nielsen (2016).”
MethodVenous air embolism during semi‐sitting craniotomy evokes thrombocytopenia
Schäfer, S., Sandalcioglu, I., Stegen, B., Neumann, A., Asgari, S., & Peters, J. (2011) · Anaesthesia · doi:10.1111/j.1365-2044.2010.06584.x · Cited by 21
Twabi, H., Manda, S., & Small, D. S. (2021) · Frontiers in Nutrition · doi:10.3389/fnut.2021.714232 · Cited by 20
Ridge rerandomization: An experimental design strategy in the presence of covariate collinearity
Branson, Z., & Shao, S. (2021) · Journal of Statistical Planning and Inference · doi:10.1016/j.jspi.2020.07.002 · Cited by 20
Hybrid Coronary Revascularization Versus Conventional Coronary Artery Bypass Surgery
Hannan, E., Wu, Y., Cozzens, K., Sundt, T., Girardi, L., Chikwe, J., Wechsler, A., Smith, C. R., Gold, J., Lahey, S., et al. (2020) · Circulation. Cardiovascular Interventions · doi:10.1161/circinterventions.120.009386 · Cited by 20
Godino, C., Pivato, C., Chiarito, M., Donahue, M., Testa, L., Colantonio, R., Cappelletti, A., Milazzo, D., Parisi, R., Nicolino, A., et al. (2017) · International Journal of Cardiology · doi:10.1016/j.ijcard.2017.06.028 · Cited by 20
Fullerton, B., Pöhlmann, B., Krohn, R., Adams, J., Gerlach, F. M., & Erler, A. (2016) · Health Services Research · doi:10.1111/1475-6773.12452 · Cited by 25
Jung, S.-Y., Sohn, H., Park, E.-J., Suh, H. S., Park, J.-W., & Kwon, J. (2016) · PLoS ONE · doi:10.1371/journal.pone.0150531 · Cited by 20
“For matching, the greedy method was used to identify these controls[25].”
MethodLim, S., Marcus, S., Singh, T., Harris, T., & Seligson, A. L. (2014) · PLoS ONE · doi:10.1371/journal.pone.0109112 · Cited by 20
“N:N), which is more efficient in balancing covariates than matching restricted to one-to-one pairs [5,7].”
BackgroundSadatsafavi, M., Lynd, L., Marra, C., Bedouch, P., & Fitzgerald, M. (2013) · Journal of Allergy and Clinical Immunology · doi:10.1016/j.jaci.2013.02.007 · Cited by 20
Economic Effects of Investment Support of Adding Value to Food Products
Mezera, J., & Špička, J. (2013) · doi:10.22004/ag.econ.148102 · Cited by 20
“Gu and Rosenbaum (1993) compared the imbalance of Mahalanobis distance metrics versus the propensity score difference in optimal 1:1 matching for numbers of covariates (P) between 2 and 20 and control/ treatment subject ratios between 2 and 6.”
BackgroundPender, J., & Reeder, R. (2010) · RePEc: Research Papers in Economics · doi:10.22004/ag.econ.262240 · Cited by 20
Constructing Matched Sets and Strata
Rosenbaum, P. (1995) · Springer series in statistics · doi:10.1007/978-1-4757-2443-1_9 · Cited by 20
Self-citationImpacts of disability on poverty: Quasi-experimental evidence from landmine amputees in Cambodia
Takasaki, Y. (2020) · Journal of Economic Behavior & Organization · doi:10.1016/j.jebo.2020.09.027 · Cited by 23
On the link between rail transit and spatial income segregation
Nilsson, I., & Delmelle, E. (2020) · Applied Geography · doi:10.1016/j.apgeog.2020.102364 · Cited by 19
Matching methods in precision oncology: An introduction and illustrative example
Weymann, D., Laskin, J., Jones, S. J. M., Lim, H., Renouf, D., Roscoe, R., Schrader, K., Sun, S., Yip, S., Marra, M., et al. (2020) · Molecular Genetics & Genomic Medicine · doi:10.1002/mgg3.1554 · Cited by 19
“Compared to nearest neighbor matching, optimal matching produces more closely matched pairs but does not improve overall balance across matched cohorts (Austin, 2014; Gu & Rosenbaum, 1993).”
BackgroundSereno, M., He, Z., Smith, C., Baena, J., Das, M., Hastings, R., Rake, G., Fennell, D., Nakas, A., Moore, D., et al. (2020) · Histopathology · doi:10.1111/his.14301 · Cited by 19
Investigating Causal DIF via Propensity Score Methods.
Liu, Y., Zumbo, B., Gustafson, P., Huang, Y., Kroc, E., & Wu, A. (2016) · Scholarworks (University of Massachusetts Amherst) · doi:10.7275/ewqz-n963 · Cited by 19
“For example, Gu and Rosenbaum (1993) conducted a simulation study, Marcus (2000) applied it to the evaluation of the Head Start compensatory education program, and Hansen and Klopfer (2006) extended the full optimal matching using optmatch R package (Hansen, 2004; Hansen, Fredrickson, Buckner, Errickson, & Solenberger…”
MethodDoes knee replacement surgery for osteoarthritis improve survival? The jury is still out
Misra, D., Lu, N., Felson, D., Choi, H. K., Seeger, J., Einhorn, T., Neogi, T., & Zhang, Y.-Q. (2016) · Annals of the Rheumatic Diseases · doi:10.1136/annrheumdis-2016-209167 · Cited by 19
Tumlinson, S. E., Sass, D. A., & Cano, S. M. (2014) · Journal of Pediatric Psychology · doi:10.1093/jpepsy/jst143 · Cited by 19
“Interested readers can consult Guo and Fraser (2010) and references cited therein for more information regarding variations of matching algorithms, sensitivity analyses, and other issues related to PSM.”
BackgroundRud, I., van Klaveren, C., Groot, W. N. J., & van den Brink, H. M. (2013) · Economics of Education Review · doi:10.1016/j.econedurev.2013.11.006 · Cited by 21
“Additionally, Mahalanobis matching is considered as a reliable approach when the vector of matching characteristics is not highly dimensional (Gu and Rosenbaum 1993; Rubin and Thomas 2000).”
A Comparison of Long-Term Mortality for Off-Pump and On-Pump Coronary Artery Bypass Graft Surgery
Wu, C., Camacho, F. T., Culliford, A., Gold, J., Wechsler, A., Higgins, R., Lahey, S., Smith, C. R., Jordan, D., & Hannan, E. (2012) · Circulation. Cardiovascular Quality and Outcomes · doi:10.1161/circoutcomes.111.963124 · Cited by 19
Ryoo, J., Hong, S., Bart, W., Shin, J., & Bradshaw, C. P. (2018) · Psychology in the Schools · doi:10.1002/pits.22134 · Cited by 18
Financial Difficulty Effects on Depressive Symptoms Among Dementia Patient Caregivers
Nam, I. (2016) · Community mental health journal · doi:10.1007/s10597-016-0033-3 · Cited by 18
“Matched pairs were removed from the pool, then the second caregiver with financial difficulty was connected to the nearest remaining caregiver without any financial difficulty, and so on (Gu and Rosenbaum 1993).”
BackgroundSecular changes in body dimensions of Royal Australian Air Force aircrew (1971–2005)
Tomkinson, G., Clark, A. J., & Blanchonette, P. (2010) · Ergonomics · doi:10.1080/00140139.2010.489967 · Cited by 18
“Using a greedy matching procedure (Gu and Rosenbaum 1993), subjects in the 1971 and 2005 datasets were matched one-for-one for decimal age, yielding 220 matched pairs.”
MethodTakeuchi, M., Yoshida, S., Kawakami, C., Kawakami, K., & Ito, S. (2022) · PLoS ONE · doi:10.1371/journal.pone.0265648 · Cited by 17
Beta-Blockers, Tachycardia and Survival Following Sepsis- An Observational Cohort Study.
Guz, D., Buchritz, S., Guz, A., Ikan, A., Babich, T., Daitch, V., Gafter-Gvili, A., Leibovici, L., & Avni, T. (2021) · Clinical Infectious Diseases · doi:10.1093/cid/ciab034 · Cited by 17
La, A., Nadarajah, V., Jauregui, J. J., Shield, W. P., Medina, S. H., Dubina, A. G., Meredith, S. J., Packer, J. D., & Henn, R. F. (2020) · Journal of Clinical Orthopaedics and Trauma · doi:10.1016/j.jcot.2019.08.009 · Cited by 17
Evaluating Propensity Score Methods in a Quasi-Experimental Study of the Impact of Menu-Labeling
Mayne, S. L., Lee, B. K., & Auchincloss, A. H. (2015) · PLoS ONE · doi:10.1371/journal.pone.0144962 · Cited by 17
“Full matching also makes use of all available data (does not drop any individuals from the analysis), and has been found to perform better than k:1 matching in simulated and real data.[19, 32] Our finding that constrained full matching resulted in an overall reduction in standardized bias is consistent with a prior st…”
ResultKretschmann, J., Vock, M., & Lüdtke, O. (2014) · Journal of Educational Psychology · doi:10.1037/a0036631 · Cited by 22
Beard, E., Aveyard, P. N., Brown, J., & West, R. (2012) · Drug and Alcohol Dependence · doi:10.1016/j.drugalcdep.2012.05.039 · Cited by 14
“In some cases, the requirement of xact matches often leads to many individuals not being matched, hich can result in larger bias than if the matches are inexact but ore individuals remain in the analysis (Rosenbaum and Rubin, 985); while Mahalanobis distance does not perform well when the ovariates are not normally di…”
BackgroundLiraglutide use and evaluation of pancreatic outcomes in a US commercially insured population
Funch, D., Mortimer, K. M., Ziyadeh, N., Seeger, J., Li, L., Norman, H., Major-Pedersen, A., Bosch-Traberg, H., Gydesen, H., & Dore, D. (2019) · Diabetes, obesity and metabolism · doi:10.1111/dom.13739 · Cited by 16
“To account for changes in prescribing patterns and availability of ADs on the market, PS estimation and matching were performed within calendar periods for each comparator.(25) For the first three years, cohorts were created by calendar quarter and annually thereafter for the remaining recruitment time.”
MethodDeming, R., Ford, M. M., Moore, M. S., Lim, S., Perumalswami, P., Weiss, J. J., Wyatt, B., Shukla, S., Litwin, A. H., Reynoso, S., et al. (2018) · Journal of Viral Hepatitis · doi:10.1111/jvh.12929 · Cited by 16
Arıkan, S., van de Vijver, F., & Yağmur, K. (2018) · doi:10.21890/ijres.382936 · Cited by 16
“Therefore, nearest neighbor matching could be used when the aim is to create well-matched pairs, whereas optimal matching could give better results when the aim is to create well-matched groups (Gu & Rosenbaum, 1993).”
MethodThe Effect of Propensity Score Matching Method on the Quantity and Quality of Matches
Jacovidis, J., Foelber, K. J., & Horst, S. J. (2016) · The Journal of Experimental Education · doi:10.1080/00220973.2016.1250209 · Cited by 16
“When choosing a matching method, consider the following: (1) the algo-rithm that is employed, (2) the structure of the matched sets (i.e., the ratio of matched comparison group members to treatment group members; Gu & Rosenbaum, 1993), and (3) whether to match with or without replacement.”
MethodBuck, D., Soden, P., Deery, S., Zettervall, S., Ultee, K., Landon, B. E., O'Malley, A., & Schermerhorn, M. (2017) · Annals of Vascular Surgery · doi:10.1016/j.avsg.2017.08.021 · Cited by 16
Celestre, P. C., Carreon, L., Lenke, L., Sucato, D., & Glassman, S. (2013) · Spine Deformity · doi:10.1016/j.jspd.2015.05.002 · Cited by 16
“Matched pairs were then created using optimal [14,15] matching without replacement using nearest-neighbor matching [16].”
MethodSadatsafavi, M., Fitzgerald, M., Marra, C., & Lynd, L. (2013) · Chest · doi:10.1378/chest.12-2773 · Cited by 16
Scherhag, C., & Boenigk, S. (2013) · Nonprofit Management and Leadership · doi:10.1002/nml.21074 · Cited by 16
TAX LIMIT REPEAL AND SCHOOL SPENDING
Nguyen-Hoang, P. (2013) · National tax journal · doi:10.17310/ntj.2013.1.05 · Cited by 16
“This study employs the full (optimal) matching algorithm introduced in Rosenbaum (1991), because full matching has been proven to perform better than other matching algorithms in minimizing variance and bias (Gu and Rosenbaum, 1993; Hansen, 2004).”
MethodKum, H., & Masterson, T. (2008) · SSRN Electronic Journal · doi:10.2139/ssrn.1136363 · Cited by 16
A quantitative causal analysis and optimization framework for inclusions of steel products
Wu, Y., Zhang, H.-D., Jian, L.-S., & Lv, Z.-M. (2024) · Advanced Engineering Informatics · doi:10.1016/j.aei.2024.102629 · Cited by 16
al-Ozairi, E., Steenackers, N., Pazmino, S., Alattar, A. T., Kandari, J. A., Almeda-Valdés, P., Antonio-Villa, N., Delfin, C., Faradji, R., Aili García-Tuomola, A., et al. (2024) · EClinicalMedicine · doi:10.1016/j.eclinm.2024.102869 · Cited by 15
“22,23 To further reduce the distance between pairs, exact matching was chosen for sex.”
Metadata
- Authors
- Gu, X. S., & Rosenbaum, P. R.
- Affiliations
- C.S. First Boston Hong Kong, Ltd.
- University of Pennsylvania (US)
- ORCID
- Rosenbaum, Paul R.
- Venue
- Journal of Computational and Graphical Statistics
- Publisher
- Informa UK Limited
- Published
- 1993-12
- Volume
- 2
- Issue
- 4
- Pages
- 405-420
- ISSN
- 1061-8600 (ISSN-L)
- 1537-2715
- OpenAlex
- W2030614017
- Cited by
- 702
- References
- 0
- Citations by year
- 2026: 20 · 2025: 29 · 2024: 27 · 2023: 32 · 2022: 35 · 2021: 41 · 2020: 46 · 2019: 37 · 2018: 51 · 2017: 36 · 2016: 36 · 2015: 40 · 2014: 36 · 2013: 35 · 2012: 27
- FWCI
- 1.01
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Journal of Computational and Graphical Statistics (publishedVersion)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
- Social Sciences › Decision Sciences › Management Science and Operations Research › Optimal Experimental Design Methods
- Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Clinical Trials
- Keywords
- optimal matching, greedy matching, pair matching, covariate balance, distance measures, observational studies
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review