Consequences of Not Accounting for One-Group Clustering in Meta-Analysis

Autor/inn/en	Citkowicz, Martyna; Polanin, Joshua R.
Institution	Society for Research on Educational Effectiveness (SREE)
Titel	Consequences of Not Accounting for One-Group Clustering in Meta-Analysis
Quelle	(2014), (14 Seiten) PDF als Volltext kostenfreie Datei Verfügbarkeit
Sprache	englisch
Dokumenttyp	gedruckt; online; Monographie
Schlagwörter	Meta Analysis; Accountability; Cluster Grouping; Effect Size; Research Design; Error of Measurement; Bias; Experimenter Characteristics; Control Groups; Experimental Groups; Research Methodology; Research Problems; Robustness (Statistics) + Suchen Sie Ihr Suchwort? Meta-analysis; Metaanalyse; Verantwortung; Eingruppierung; Forschungsdesign; Messfehler; Research method; Forschungsmethode; Forschungskritik; Widerstandsfähigkeit
Abstract	Meta-analyses are syntheses of effect-size estimates obtained from a collection of studies to summarize a particular field or topic (Hedges, 1992; Lipsey & Wilson, 2001). These reviews are used to integrate knowledge that can inform both scientific inquiry and public policy, therefore it is important to ensure that the estimates of the effect sizes are both robust and free of bias. One such scenario under which bias can occur is from a failure to account for clustering. Clustering occurs when, for example, schools are assigned to conditions and students are grouped together within each school. Occasionally, study designs are such that clustering exists in one condition but not the other. These studies may be referred to as one-group cluster-randomized designs or partially clustered designs. Despite recent advances, researchers do not yet have an understanding of the degree to which "not" accounting for clustering will bias the overall meta-analytic mean effect and its variance. The purpose of this study is to investigate the consequences of not accounting for one-group clustering in meta-analyses. The authors conducted a series of simulations in order to find out how distorted meta-analytic results can be when analysts fail to correct effect sizes for one-group clustering. The results of these analyses, in addition, demonstrate the extent to which researchers may be confident in previously conducted meta-analyses that did not have the tools to adjust the effects and variance in such designs. 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