Multivariate meta‐analysis: the effect of ignoring within‐study correlation

By R. Riley. 2009

213 citations recorded by the source corpus.

Subfields: Economics and Econometrics, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/88058f99531a043317b2fc67d7354f1615e6b97e

How researchers see this paper

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

In Economics and Econometrics

In Statistics and Probability

Citations in the corpus

Multivariate meta-analysis: Potential and promise (2011, methodology): “Riley [13] notes that, with the exception of diagnostic test studies, ‘multivariate meta-analysis methods are rarely used by practitioners in systematic reviews’.”
Multivariate meta-analysis for non-linear and other multi-parameter associations (2012, unknown): “Methods to deal with missing correlations have been developed [50,51], although this issue needs to be explored further, especially for multi-parameter associations.”
Rejoinder to commentaries on ‘Multivariate meta‐analysis: Potential and promise’ (2011, methodology): “Another advantage of the multivariate methods that we consider especially important is the ‘borrowing of strength’ that we refer to in our article and we thank David Cox for his three (extreme) cases that clarify the issues involved, echoing analytic examples elsewhere [2, 3].”
Meta‐analysis using individual participant data: one‐stage and two‐stage approaches, and why they may differ (2016, unknown): “These patients responses, and indeed subsequent parameter estimates (such as treatment effects at multiple time points), will be correlated, and ignoring this in the meta-analysis is inadvisable [84,85].”
Career Adaptability and Career Decision Self-Efficacy: Meta-Analysis (2021, methodology): “There are several theoretical advantages of multivariate modeling (Riley, 2009), but there are practical limitations. Overfitting a multivariate meta-regression model occurs when the number of studies is small with respect to the number of moderators. This is particularly problematic when categorica”
Multivariate Random-effects Meta-regression: Updates to Mvmeta (2011, methodology): “wscorr(riley) can be used when within-study correlations are unknown. It uses the alternative model of Riley, Thompson, and Abrams (2008) to estimate an overall correlation; see section 3.5. Riley (2009) discusses other ways to handle unknown within-study correlations.”
Meta‐analysis of variation: ecological and evolutionary applications and beyond (2015, unknown): “A difficulty of such a bivariatemeta-analytic model is that we do not know sampling (error) covariance (Riley 2009), although we note some solutions to this issue have been proposed (Riley, Thompson&Abrams 2008).”
Robust variance estimation in meta‐regression with dependent effect size estimates (2010, background): “Our experience is also that these estimates are not highly sensitive to choices of covariance structure in practical situations, but it does seem reasonable that estimates of between-study variance components will be most sensitive to within-study correlation structure when between-study covariance ”
A Guide to Conducting a Meta-Analysis with Non-Independent Effect Sizes (2019, background): “Several authors (e.g., Demidenko, 2013; Riley, 2009) have shown that the estimated fixed effects in a multivariate meta-analysis usually have smaller SEs.”