A comparison of statistical methods for meta-analysis

Department of Mathematics and Statistics, Richard Berry Building, The University of Melbourne, Victoria 3010, Australia.
Statistics in Medicine (Impact Factor: 1.83). 03/2001; 20(6):825-40. DOI: 10.1002/sim.650
Source: PubMed


Meta-analysis may be used to estimate an overall effect across a number of similar studies. A number of statistical techniques are currently used to combine individual study results. The simplest of these is based on a fixed effects model, which assumes the true effect is the same for all studies. A random effects model, however, allows the true effect to vary across studies, with the mean true effect the parameter of interest. We consider three methods currently used for estimation within the framework of a random effects model, and illustrate them by applying each method to a collection of six studies on the effect of aspirin after myocardial infarction. These methods are compared using estimated coverage probabilities of confidence intervals for the overall effect. The techniques considered all generally have coverages below the nominal level, and in particular it is shown that the commonly used DerSimonian and Laird method does not adequately reflect the error associated with parameter estimation, especially when the number of studies is small.

Download full-text


Available from: Ian Robert Gordon, Sep 19, 2014
  • Source
    • "Although fixed effect models are widely used in two-stage meta-analyses, even when heterogeneity is not zero (Kontopantelis, Springate, and Reeves 2013), accounting for even low levels of between-cluster variability is a more conservative approach (Hunter and Schmidt 2000). When a fixed effects model is incorrectly assumed, both coverage and power deteriorate as true heterogeneity increases (Brockwell and Gordon 2001; Kontopantelis and Reeves 2012). Analogously, for patient data analyses, we would expect poor fit from model(1), the fixed effect approach, in the presence of heterogeneity. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Simulations are a practical and reliable approach to power calculations, especially for multi-level mixed effects models where the analytic solutions can be very complex. In addition, power calculations are model-specific and multi-level mixed effects models are defined by a plethora of parameters. In other words, model variations in this context are numerous and so are the tailored algebraic calculations. This article describes ipdpower in Stata, a new simulations-based command that calculates power for mixed effects two-level data structures. Although the command was developed having individual patient data meta-analyses and primary care databases analyses in mind, where patients are nested within studies and general practices respectively, the methods apply to any two-level structure.
    • "That is, each study is considered to estimate its own population effect size (Hedges 1994; Hedges and Olkin 1985). Moreover, this model is recommended when the number of studies is small (Brockwell and Gordon 2011). To estimate the heterogeneity, the Q statistic was used, completed with the I 2 statistic. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The present study is the first meta-analytic study about the prevalence of intimate partner violence (IPV) in self-identified lesbians in same-sex couples. It summarizes the scientific evidence from studies published from 1990 to 2013. First, 1,184 studies were identified, then 59 studies were pre-selected, and finally 14 studies were chosen that met the criteria for inclusion and methodological quality. The studies were conducted in the USA, using non-probabilistic sampling methods, and they were characterized by their high level of heterogeneity. The mean prevalence of victimization in IPV over the lifespan is 48 % (95 % CI, 44–52 %) and 15 % (95 % CI, 5–30 %) in the current/most recent relationship, with the difference being statistically significant between over the lifespan and current/most recent relationship IPV. The mean prevalence of victimization in physical violence over the lifespan is 18 % (95 % CI, 0–48 %), in sexual violence 14 % (95 % CI, 0–37 %), and in psychological/emotional violence 43 % (95 % CI, 14–73 %). The high prevalence suggests the need to implement IPV prevention programs among lesbians, as well as homophobia prevention programs. Moreover, the methodological quality of prevalence studies should be improved. The limited number of studies considered in each thematic block and the high heterogeneity of their results should be taken into account.
    Sexuality Research and Social Policy: Journal of NSRC 03/2015; 12(1). DOI:10.1007/s13178-014-0164-7 · 0.87 Impact Factor
  • Source
    • "First, this study was not able to exhaust all previous studies on destination image and tourist loyalty , although FSN demonstrated a high level of validity. Second, meta-analysis has also been criticized for having to lose contextual information such as characteristics of the samples and variations in the design quality (Brockwell & Gordon, 2001; Field, 2003). Similar to other meta-analytic studies, the paper was unable to report every inter-study differences, sample characteristics, and variation in model specification. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Extant literature is inconclusive on the linkage between destination image and tourist loyalty, due to the multi-dimensional nature of the two concepts. The present study attempts to draw some informative conclusions about the relationship through a meta-analysis. A research framework was proposed in which 14 hypotheses were developed. A total of 66 independent studies were synthesized and analyzed. The findings reveal that the impact of destination image on tourist loyalty is significant, with varying degrees. Specifically, overall image has the greatest impact on tourist loyalty, followed by affective image and cognitive image. Cognitive affective joint image fails to demonstrate a stable impact on tourist loyalty. Of the three levels of tourist loyalty, destination image has the greatest impact on composite loyalty, and then on attitudinal loyalty and behavioral loyalty, successively. The findings are discussed in light of their theoretical and practical implications for destination marketing and management.
    Tourism Management 02/2014; 40:213-223. DOI:10.1016/j.tourman.2013.06.006 · 2.57 Impact Factor
Show more