funnel plot

funnel plot

a graphic method of detecting publication bias. The estimate of risk derived from a set of epidemiologic studies used in a metaanalysis is plotted against sample size. If there is no publication bias, the plot is funnel-shaped; if studies giving significant results are more likely to be published than negative studies, the plot is asymmetric.
See also: metaanalysis.
A scatterplot of treatment effect against a measure of study size used primarily as a visual aid for detecting bias or systematic heterogeneity. A symmetric inverted funnel shape is seen in ‘well-behaved’ data for which publication bias is unlikely and which is thus suited for inclusion in a meta-analysis. An asymmetric funnel indicates a relationship between treatment effect and study size, due either to publication bias or a systematic difference between smaller and larger studies—‘small study effects’

fun·nel plot

(fŭn'ĕl plot)
A graphic method of detecting publication bias. The estimate of risk derived from a set of epidemiologic studies used in a metaanalysis is plotted against sample size. If there is no publication bias, the plot is funnel shaped; if studies giving significant results are more likely to be published than negative studies, the plot is asymmetric.
See also: metaanalysis
References in periodicals archive ?
Visual inspection of the funnel plot did not demonstrate the asymmetry typically associated with publication bias (Fig.
12) Publication bias was assessed by constructing a funnel plot.
For all series, we present HSMRs for patients discharged in the financial year 2008/2009 by summarizing the differences between regular and the nine sets of variant HSMRs in bands of < 5, 5-9, and 10 + and in terms of change in funnel plot outlier status using 99.
To investigate publication bias, a funnel plot was generated, using the datasets from the included clinical trials.
5], the estimate accounting for funnel plot asymmetry was 1.
In addition, we conducted trim-and-fill analyses, which, in the presence of funnel plot asymmetry, recalculate the meta-estimate based on hypothetical, unobserved studies that would have been necessary to create a symmetrical funnel plot (Duval and Tweedie 2000; Higgins and Green 2008).
Lemmers was aware of the Scottish study and said a funnel plot they performed showed "the exact same results" between older and newer studies.
The funnel plot was asymmetric in favor of smaller trials with greater effects (Egger's P = .
To assess the degree to which publication bias may have influenced the results, a funnel plot was computed along with the associated Begg and Mazumdar's rank correlation and Egger's linear regression tests (Borenstein, 2005).
As data for all surgeons fall within predefined limits on a funnel plot, however, we would argue that there would be little basis for using the data presented here to select one, rather than another, surgeon in this list.
Publication bias was evaluated through funnel plots and Egger's linear regression test, with six studies demonstrating no publication bias.