Combining Propensity Score Methods and Complex Survey Data to Estimate Population Treatment Effects

Autor/inn/en	Stuart, Elizabeth A.; Dong, Nianbo; Lenis, David
Institution	Society for Research on Educational Effectiveness (SREE)
Titel	Combining Propensity Score Methods and Complex Survey Data to Estimate Population Treatment Effects
Quelle	(2016), (11 Seiten) PDF als Volltext kostenfreie Datei Verfügbarkeit
Sprache	englisch
Dokumenttyp	gedruckt; online; Monographie
Schlagwörter	Outcomes of Treatment; Probability; Surveys; Computation; Statistical Analysis; Children; Longitudinal Studies; Sampling; Grants; Kindergarten; Mathematics Achievement; Early Childhood Longitudinal Survey + Suchen Sie Ihr Suchwort? Wahrscheinlichkeitsrechnung; Wahrscheinlichkeitstheorie; Survey; Umfrage; Befragung; Statistische Analyse; Child; Kind; Kinder; Longitudinal study; Longitudinal method; Longitudinal methods; Längsschnittuntersuchung; Grant; Finanzielle Beihilfe; Mathmatics sikills; Mathmatics achievement; Mathematical ability; Mathematische Kompetenz
Abstract	Complex surveys are often used to estimate causal effects regarding the effects of interventions or exposures of interest. Propensity scores (Rosenbaum & Rubin, 1983) have emerged as one popular and effective tool for causal inference in non-experimental studies, as they can help ensure that groups being compared are similar with respect to a large set of observed characteristics. However, little work has investigated how best to combine propensity scores and complex survey data to estimate population treatment effects. This has led to confusion in the literature, with many applied researchers using inappropriate methods or claiming representativeness of study results when the analysis does not warrant such claims (DuGoff, Schuler and Stuart, 2014; Ridgeway, Kovalchik, Griffin, and Kabeto, 2015). One way to think about the complication of estimating population treatment effects using data from a complex survey is that when there are heterogeneities in treatment assignment, sample selection probabilities, and treatment effects, failure to take into account sampling weights might cause biased population treatment effect estimates. Ignoring sampling weights leads mainly to external validity bias, which occurs when people inappropriately make inferences from the unrepresentative analytic sample to the target population. This work aims to clarify the results and recommendations regarding the use of propensity scores with complex survey data. Tables are appended. (ERIC).
Anmerkungen	Society for Research on Educational Effectiveness. 2040 Sheridan Road, Evanston, IL 60208. Tel: 202-495-0920; Fax: 202-640-4401; e-mail: inquiries@sree.org; Web site: http://www.sree.org
Erfasst von	ERIC (Education Resources Information Center), Washington, DC
Update	2020/1/01