Hi there im doing a meta-analysis on review manager.
I know how to do a funnel plot, I think there is a way to do a statistics to further examine this. I can't work out how to do it despite looking at manual etc! Any help much appreciated please
Sorry, I have no idea. I do apologise. All best with the work,
Terry Langford.
Thanks anyway think ive worked it out you can calculate standard error and then correlate with effect size....
I think reading manuals is not a bad thing, although some of them are very unclear.
Most of the publication bias assessments tell you whether there is asymmetry in the funnel plot or not. That sounds like what you have figured out how to do. On the other hand, only Duval and Tweedie's trim & fill attempts to estimate the magnitude of the publication bias. In my opinion, the trim & fill correction to the overall effect size is usually the preferred way to go. However, there are a couple situations in which the trim & fill correction doesn't work properly (e.g., when heterogeneity is high).
Is there a way to calculate an actual number around publication bias on review manager rather than just a funnel plot? Thanks
Thomas, to my knowledge RevMan does not offer any statistical tools (e.g. Egger's test) for calculating publication bias. The only thing you can do with RevMan is visually inspect the funnel plot for asymmetry. If you want to use statistical methods, you will need to use other software (e.g. R, STATA, SPSS).
Hi Spyros,
Could you tell me how to do eggers test in SPSS? I'm struggling to find a guide online?
Hi all! Maybe this would be helpful:
https://www.meta-analysis.com/downloads/PBSoftware.pdf
Hi Thomas. I have the sama problem as you. But I suggest you use Jamovi, a statistical software based on R. It is very user-friendly and can calculate publication bias automatically. Hope this would be helpful.
https://www.jamovi.org/download.html
Egger’s linear regression method (Egger et al., 1997; Sterne et al., 2001b, c), like
the rank correlation test, is intended to quantify the bias captured by the funnel
plot. This differs from Begg and Mazumdar’s test in that Egger uses the actual
values of the effect sizes and their precision, rather than ranks. In the Egger test,
the standard normal deviate is regressed on precision, defined as the inverse of
the standard error. The intercept in this regression corresponds to the slope in a
weighted regression of the effect size on the standard error. As was true for the
rank correlation test, the significance test should be two-tailed. I recommend this link: https://www.meta-analysis.com/downloads/PBSoftware.pdf
I think revman cannot calculate publication bias . It is better to conduct publication bias on state.
In my opinion, We can measure publication bias by comparing the results of published and unpublished studies addressing the same question
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Social Media