Performance of statistical methods for meta-analysis when true study effects are non-normally distributed: A simulation study
Open accessThe DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index.
Abstract
Meta-analysis (MA) is a statistical methodology that combines the results of several independent studies considered by the analyst to be ‘combinable’. The simplest approach, the fixed-effects (FE) model, assumes the true effect to be the same in all studies, while the random-effects (RE) family of models allows the true effect to vary across studies. However, all methods are only correct asymptotically, while some RE models assume that the true effects are normally distributed. In practice, MA methods are frequently applied when study numbers are small and the normality of the effect distribution unknown or unlikely. In this article, we discuss the performance of the FE approach and seven frequentist RE MA methods: DerSimonian–Laird, Q-based, maximum likelihood, profile likelihood, Biggerstaff–Tweedie, Sidik–Jonkman and Follmann–Proschan. We covered numerous scenarios by varying the MA sizes (small to moderate), the degree of heterogeneity (zero to very large) and the distribution of the effect sizes (normal, skew-normal and ‘extremely’ non-normal). Performance was evaluated in terms of coverage (Type I error), power (Type II error) and overall effect estimation (accuracy of point estimates and error intervals).
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Meta-analysis in clinical trials
DerSimonian, R., & Laird, N. (1986) · Controlled Clinical Trials · doi:10.1016/0197-2456(86)90046-2 · Cited by 39,552
Quantifying heterogeneity in a meta‐analysis
Higgins, J. P. T., & Thompson, S. G. (2002) · Statistics in Medicine · doi:10.1002/sim.1186 · Cited by 36,763
Primary, Secondary, and Meta-Analysis of Research
GLASS, G. V. (1976) · Educational Researcher · doi:10.3102/0013189x005010003 · Cited by 4,890 · cited 1×
“The progenitor of this label, an educational researcher, defined MA as ‘the statistical analysis of a large collection of analysis results from individual studies for the purpose of integrating the findings’.(2) A primary concern for meta-analysts is the statistical heterogeneity between the studies included in an ana…”
BackgroundThe unicorn, the normal curve, and other improbable creatures.
Micceri, T. (1989) · Psychological Bulletin · doi:10.1037//0033-2909.105.1.156 · Cited by 1,485
Systematic Review: Why sources of heterogeneity in meta-analysis should be investigated
Thompson, S. G. (1994) · BMJ · doi:10.1136/bmj.309.6965.1351 · Cited by 1,017 · cited 3×
“Nevertheless, critics argue that it may still be inappropriate to combine heterogeneous trials: an ‘average’ MA result may not be clinically useful if the effect is context-specific.(4,14) An additional problem for both RE and FE models is that most commonly available analysis techniques assume true study effects to b…”
BackgroundNonparametric maximum likelihood estimation of a mixing distribution
Laird, N. (1978) · Journal of the American Statistical Association · doi:10.1080/01621459.1978.10480103 · Cited by 753
The power of statistical tests in meta-analysis
Hedges, L. V., & Pigott, T. D. (2001) · Psychological Methods · doi:10.1037/1082-989x.6.3.203 · Cited by 716 · cited 1×
“2 Power probability A power comparison of the various models is important if we are to evaluate their performance in every respect.(33) Overton(34) compared the FE to the RE model using both Type I error rates and power probabilities.”
MethodMeta-analysis: Potentials and promise
Egger, M., & Smith, G. D. (1997) · BMJ · doi:10.1136/bmj.315.7119.1371 · Cited by 646
A comparison of statistical methods for meta‐analysis
Brockwell, S. E., & Gordon, I. R. (2001) · Statistics in Medicine · doi:10.1002/sim.650 · Cited by 573
A LIKELIHOOD APPROACH TO META-ANALYSIS WITH RANDOM EFFECTS
HARDY, R. J., & THOMPSON, S. G. (1996) · Statistics in Medicine · doi:10.1002/(sici)1097-0258(19960330)15:6<619::aid-sim188>3.0.co;2-a · Cited by 522
Thompson, S. G., & Pocock, S. J. (1991) · The Lancet · doi:10.1016/0140-6736(91)91975-z · Cited by 487
Approaches to heterogeneity in meta‐analysis
Petitti, D. B. (2001) · Statistics in Medicine · doi:10.1002/sim.1091 · Cited by 464
Problems Arising in the Analysis of a Series of Similar Experiments
Cochran, W. G. (1937) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.2307/2984123 · Cited by 448
A simple confidence interval for meta-analysis
Sidik, K., & Jonkman, J. N. (2002) · Statistics in Medicine · doi:10.1002/sim.1262 · Cited by 431 · cited 1×
“A solution to the coverage deterioration problem of the DL model, for small numbers of studies, was later proposed by Sidik and Jonkman (SJ).(20) This model is a non-iterative variant of the DL model, based on the t-distribution, with higher coverage probabilities than the DL method, especially for small numbers of st…”
MethodA Probability Distribution and its Uses in Fitting Data
Ramberg, J. S., Dudewicz, E. J., Tadikamalla, P. R., & Mykytka, E. F. (1979) · Technometrics · doi:10.1080/00401706.1979.10489750 · Cited by 426
A comparison of heterogeneity variance estimators in combining results of studies
Sidik, K., & Jonkman, J. N. (2007) · Statistics in Medicine · doi:10.1002/sim.2688 · Cited by 380
Evaluation of old and new tests of heterogeneity in epidemiologic meta-analysis
Takkouche, B., Cadarso-Suarez, C., & Spiegelman, D. (1999) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a009981 · Cited by 380
Overton, R. C. (1998) · Psychological Methods · doi:10.1037//1082-989x.3.3.354 · Cited by 356
Random-Effects Meta-Analyses Are Not Always Conservative
Poole, C., & Greenland, S. (1999) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a010035 · Cited by 338
INCORPORATING VARIABILITY IN ESTIMATES OF HETEROGENEITY IN THE RANDOM EFFECTS MODEL IN META-ANALYSIS
BIGGERSTAFF, B. J., & TWEEDIE, R. L. (1997) · Statistics in Medicine · doi:10.1002/(sici)1097-0258(19970415)16:7<753::aid-sim494>3.0.co;2-g · Cited by 272
Methods for Exploring Heterogeneity in Meta-Analysis
Song, F., Sheldon, T. A., Sutton, A. J., Abrams, K. R., & Jones, D. R. (2001) · Evaluation & the Health Professions · doi:10.1177/01632780122034849 · Cited by 253
Investigating causes of heterogeneity in systematic reviews
Glasziou, P. P., & Sanders, S. L. (2002) · Statistics in Medicine · doi:10.1002/sim.1183 · Cited by 213
Metaan: Random-effects Meta-analysis
Kontopantelis, E., & Reeves, D. (2010) · The Stata Journal Promoting communications on statistics and Stata · doi:10.1177/1536867x1001000307 · Cited by 161
Self-citationA simulation study comparing properties of heterogeneity measures in meta‐analyses
Mittlböck, M., & Heinzl, H. (2006) · Statistics in Medicine · doi:10.1002/sim.2692 · Cited by 148
Valid Inference in Random Effects Meta‐Analysis
Follmann, D. A., & Proschan, M. A. (1999) · Biometrics · doi:10.1111/j.0006-341x.1999.00732.x · Cited by 128
Some general points in estimating heterogeneity variance with the DerSimonian-Laird estimator
Böhning, D., Malzahn, U., Dietz, E., Schlattmann, P., Viwatwongkasem, C., & Biggeri, A. (2002) · Biostatistics · doi:10.1093/biostatistics/3.4.445 · Cited by 128
Controversies in meta-analysis: the case of the trials of serum cholesterol reduction
Thompson, S. G. (1993) · Statistical Methods in Medical Research · doi:10.1177/096228029300200205 · Cited by 118
Generating Non-normal Data for Simulation of Structural Equation Models Using Mattson's Method
Reinartz, W., Echambadi, R., & Chin, W. W. (2002) · Multivariate Behavioral Research · doi:10.1207/s15327906mbr3702_03 · Cited by 71
MetaEasy: A Meta-Analysis Add-In for MicrosoftExcel
Kontopantelis, E., & Reeves, D. (2009) · Journal of Statistical Software · doi:10.18637/jss.v030.i07 · Cited by 65
Self-citationMeta-analysis by random effect modelling in generalized linear models
Aitkin, M. (1999) · Statistics in Medicine · doi:10.1002/(sici)1097-0258(19990915/30)18:17/18<2343::aid-sim260>3.0.co;2-3 · Cited by 51
Engels, E. (2000) · doi:10.1002/1097-0258(20000715)19:13 · Cited by 2
A meta-analysis add-in for microsoft excel
Kontopantelis, E., & Reeves, D. S. (2009) · Research Explorer (The University of Manchester) · Cited by 2
Self-citationDetecting and describing heterogeneity in meta-analysis
Hardy, R. J., & Thompson, S. G. (1998) · Statistics in Medicine · doi:10.1002/(sici)1097-0258(19980430)17:8<841::aid-sim781>3.0.co;2-d
Research works printed but not linked: 7
- Higgins JPT, 2005, Cochrane handbook for systematic reviews of interventions 4.2.5 [updated May 2005]
- 10.1177/016327870102400203 · doi:10.1177/016327870102400203 · Paper page
- 10.1037/0033-2909.105.1.156 · doi:10.1037/0033-2909.105.1.156 · Paper page
- 10.1002/1097-0258(20000715)19:13<1707::AID-SIM491>3.0.CO;2-P · doi:10.1002/1097-0258(20000715)19:13<1707::aid-sim491>3.0.co;2-p
- 10.1007/BF02294081 · doi:10.1007/bf02294081
- 10.1037/1082-989X.3.3.354 · doi:10.1037/1082-989x.3.3.354 · Paper page
- The Cochrane Collaboration. Review Manager for Windows. [4.2.10]. 2007. Ref type: Computer Program.
Cited by
Meta-analysis in clinical trials revisited
DerSimonian, R., & Laird, N. (2015) · Contemporary Clinical Trials · doi:10.1016/j.cct.2015.09.002 · Cited by 2,677
IntHout, J., Ioannidis, J. P., & Borm, G. F. (2014) · BMC Medical Research Methodology · doi:10.1186/1471-2288-14-25 · Cited by 2,419
“Kontopantelis and Reeves [27] showed that with slight heterogeneity the coverage of the HKSJ method was consistently 94% when the true effects were not distributed according to the normal or t-distribution, but with”
MethodMethods to estimate the between‐study variance and its uncertainty in meta‐analysis
Veroniki, A. A., Jackson, D., Viechtbauer, W., Bender, R., Bowden, J., Knapp, G., Kuss, O., Higgins, J. P., Langan, D., & Salanti, G. (2016) · Research Synthesis Methods · doi:10.1002/jrsm.1164 · Cited by 1,685
“Several approaches have been suggested to estimate a CI around the mean effect, the performance of which has been examined in simulation studies under various scenarios (Brockwell and Gordon, 2001; Hartung, 1999; Kontopantelis and Reeves, 2012).”
A comparison of heterogeneity variance estimators in simulated random‐effects meta‐analyses
Langan, D., Higgins, J. P. T., Jackson, D., Bowden, J., Veroniki, A. A., Kontopantelis, E., Viechtbauer, W., & Simmonds, M. (2019) · Research Synthesis Methods · doi:10.1002/jrsm.1316 · Cited by 1,074
Self-citationSchwingshackl, L., & Hoffmann, G. (2015) · Journal of the Academy of Nutrition and Dietetics · doi:10.1016/j.jand.2014.12.009 · Cited by 1,022
Metadata
- Authors
- Kontopantelis, E., & Reeves, D.
- Venue
- Statistical Methods in Medical Research
- Publisher
- SAGE Publications
- Published
- 2010-12-09
- Volume
- 21
- Issue
- 4
- Pages
- 409-426
- ISSN
- 0962-2802 (ISSN-L)
- 1477-0334
- OpenAlex
- W1977947020
- Cited by
- 264
- References
- 33
- FWCI
- 2.32
- Type
- journal article
- Language
- en
- Open access
- green
- Licence
- journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
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
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.