Concept Architecture
Concept
Theoretically, Begg Test is a non-parametric statistical test used to detect publication bias in meta-analysis. It assesses whether the magnitude of study effect estimates is associated with their sampling variance, based on the assumption that, in the absence of publication bias, effect sizes should be independent of study precision. The method is founded on rank correlation and was developed as a formal statistical complement to visual inspection of funnel plots.
Mathematically, the Begg test is based on Kendall's rank correlation coefficient between the standardised treatment effects and their variances or standard errors. The test statistic evaluates whether a statistically significant monotonic association exists, with significant correlation suggesting potential publication bias or small-study effects.
In practice, the Begg test is calculated using study-level effect estimates and their variances after completion of a meta-analysis. It is commonly reported alongside Egger's regression test and funnel plots when assessing the robustness of systematic reviews and health technology assessments. Because the test has relatively low statistical power, particularly when few studies are available, it is generally interpreted together with other assessments of publication bias.
Purpose
Used to detect potential publication bias and small-study effects in meta-analysis by assessing the association between study effect estimates and study precision.
Mathematical Formulae
Primary Formula
K = Kendall's �(effect estimate rank, variance rank)
Test statistic:
Z = � / SE(�)
where:
- � = Kendall's rank correlation coefficient
- SE(�) = standard error of �
Under the null hypothesis:
Z ~ N(0,1)
Supporting Formulae
Kendall's rank correlation coefficient:
� = (C ? D) / (n(n ? 1) / 2)
where:
- C = number of concordant pairs
- D = number of discordant pairs
- n = number of studies
Related Mathematical Methods
- Kendall's Rank Correlation
- Funnel Plot Analysis
- Egger's Regression Test
- Random-Effects Meta-Analysis
- Fixed-Effect Meta-Analysis
Example
A meta-analysis combines 18 randomised controlled trials evaluating a new antihypertensive therapy. Following estimation of the pooled treatment effect, the Begg test produces Z = 0.94 with p = 0.35. Because the result is not statistically significant, there is no evidence that publication bias is detectable using the Begg test, although funnel plot inspection and Egger's regression are also performed to support the assessment.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| RANK.AVG | =RANK.AVG(A2,$A$2:$A$21) | Rank study effect estimates |
| RANK.AVG | =RANK.AVG(B2,$B$2:$B$21) | Rank study variances or standard errors |
| CORREL | =CORREL(C2:C21,D2:D21) | Preliminary assessment of rank association (not a substitute for Kendall's �) |
| NORM.S.DIST | =NORM.S.DIST(E2,TRUE) | Calculate p-values from the standard normal approximation |
VBA (Optional)
Automate publication bias assessment by calculating Begg test statistics across multiple meta-analyses and generating summary reports with accompanying funnel plots.
Sources
- Begg CB, Mazumdar M. Operating Characteristics of a Rank Correlation Test for Publication Bias. Biometrics. 1994.
- Higgins JPT, Thomas J, Chandler J, et al. Cochrane Handbook for Systematic Reviews of Interventions.
- Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Introduction to Meta-Analysis.
- NICE. Health Technology Evaluation Manual.
- ISPOR Good Practice Reports.
Related Concepts (2)
Library
Publications
1
Cochrane Handbook for Systematic Reviews of Interventions — Higgins, Thomas, Chandler, Cumpston, Li, Page & Welch, 2nd Edition ed., 2019 (John Wiley & Sons / Cochrane)
The standard guide to planning, conducting, interpreting and reporting systematic reviews of health interventions, with extensive material on meta-analysis, network meta-analysis, risk of bias, GRADE, equity, complex interventions and economics evidence. Maintained as a living online resource.
BookView source →
Frequently Asked Questions (6)
What is the Begg test?
A statistical test for detecting publication bias in a meta-analysis, based on the rank correlation between effect size and its variance.
Source: Begg & Mazumdar 1994
What does the Begg test look for in a meta-analysis?
The Begg test looks for publication bias in a meta-analysis by examining whether a study's estimated effect is correlated with how precisely it was measured. If small, imprecise studies tend to report larger effects than big, precise ones, that rank correlation suggests the smaller null results were never published. A significant correlation is taken as a warning sign that the collected studies may not represent all those conducted. Detecting a link between size and result is its aim. Begg and Mazumdar (1994) describe this test.
Source: Begg & Mazumdar 1994
How does the Begg test work?
The Begg test works by calculating the rank correlation, using a measure such as Kendall's tau, between the standardised effect estimates and their variances across the studies in a meta-analysis. A significant correlation indicates that effect size is associated with precision, consistent with funnel plot asymmetry and possible publication bias, since smaller, less precise studies show systematically different effects. The absence of correlation is consistent with no bias. So the Begg test quantifies the association between effect size and precision through rank correlation, providing a statistical assessment of the asymmetry that publication bias would produce.
Source: Begg & Mazumdar 1994
What does the Begg test detect?
The Begg test detects an association between study effect sizes and their precision, which corresponds to asymmetry in a funnel plot and can indicate publication bias, where smaller studies with unfavourable results are less likely to be published, leaving a skewed set of studies. However, such asymmetry can also arise from causes other than publication bias, such as genuine heterogeneity or study quality differences. So the Begg test detects the funnel plot asymmetry that publication bias would cause, but a positive result signals possible bias rather than confirming it, requiring cautious interpretation and consideration of other explanations.
Source: Sterne, Gavaghan & Egger 2000
What are the limitations of the Begg test?
The limitations of the Begg test include low statistical power, especially when few studies are included, so it may fail to detect real publication bias; and the fact that a significant result indicates asymmetry, which can arise from causes other than publication bias, such as heterogeneity or quality differences, so it does not confirm bias. It is generally considered less powerful than the Egger test. These limitations mean the Begg test is interpreted cautiously, used alongside other methods and visual inspection, and its result treated as suggestive of possible publication bias rather than definitive, particularly when the number of studies is small.
Source: Begg & Mazumdar 1994
How does the Begg test relate to the Egger test?
The Begg test relates to the Egger test as an alternative statistical method for assessing funnel plot asymmetry and possible publication bias: the Begg test uses the rank correlation between effect size and variance, while the Egger test uses a regression of the effect estimates against their precision. Both aim to detect the asymmetry that publication bias would produce, but the Egger test is generally considered more powerful. So the two are complementary approaches to the same question, and both are interpreted cautiously, since asymmetry can have causes other than publication bias, with the Egger test often preferred for its greater sensitivity.
Source: Begg & Mazumdar 1994
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 1 Dec 2025
Content version: 1.0.0
Canonical Identity
- Persistent URI
- https://healtheconomics.wiki/concept/begg-test
- Term code
- HE-ES-ESM-003
Stable URI · Machine-readable · Resolvable · CC BY 4.0