I had been involved in meta-analysis and a major contribution for my work is from researchgate scientists. Thank you all. I like to know how to interpret the test of Beggs and Eggers in stata for assessing publication bias?
Both tests are formalized statistical tests for assessing funnel plot asymmetry. Thus, if the tests provide a significant result, it means that the funnel plot is asymmetric. In other words, small studies (i.e., with smaller precision) show larger effect sizes. This makes it likely that publication bias has occured: small studies with insignificant findings (small effect sizes) are not published and therefore not included in your meta-analysis.
If the tests fail to detect publication bias, you can argue that it seems not to be a big threat to the validity of the research.
Both tests are formalized statistical tests for assessing funnel plot asymmetry. Thus, if the tests provide a significant result, it means that the funnel plot is asymmetric. In other words, small studies (i.e., with smaller precision) show larger effect sizes. This makes it likely that publication bias has occured: small studies with insignificant findings (small effect sizes) are not published and therefore not included in your meta-analysis.
If the tests fail to detect publication bias, you can argue that it seems not to be a big threat to the validity of the research.
Unless you have many studies in your meta-analysis, the Begg method has very low power to detect biases . Thus, use other precise test like harbord test to check for asymmetry if your outcome variable is binary.
However, if p value is significant it claims the presence of publication bias.
In stata you can use this command
metabias tcases tnoncases ccases cnoncases,or harbord graph mlabel( subgrouping variable)
It is always important to consider that both tests, when asymmetry is present, indicate rather than confirm the presence of bias. These tests consider a significance level of 10%.
If there is no 'small sample' bias across a series of studies in a meta-analysis then the estimates of effect should vary (due to random error) most with the small studies and least with the large studies. This fact lead to the use of plots of sample size against effect estimate (the original funnel plot). Bias is likely to cause asymmetry in such plots. As sample size is not the only determinant of the precision of an effect estimate, richer information for detecting bias can be gained from plotting the standard errors against their effect estimates. The reciprocal of the standard error is referred to as precision. It is common to plot effect estimates on the horizontal axis and the measure of study size on the vertical axis. This is the opposite of the usual convention for two-way plots, in which the outcome (e.g., intervention effect) is plotted on the vertical axis and the covariate (e.g., study size) is plotted on the horizontal axis.
What if the p-value is significant but the confidence interval includes 1?
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation