Forest Plots — plot.uvmeta • metamisc

Function to create forest plots for objects of class "uvmeta".

Usage

# S3 method for uvmeta
plot(x, sort = "asc", ...)

Arguments

x: An object of class "uvmeta"
sort: By default, studies are ordered by ascending effect size (sort="asc"). For study ordering by descending effect size, choose sort="desc". For any other value, study ordering is ignored.
...: Additional arguments which are passed to forest.

Details

The forest plot shows the performance estimates of each validation with corresponding confidence intervals. A polygon is added to the bottom of the forest plot, showing the summary estimate based on the model. A 95% prediction interval is added by default for random-effects models, the dotted line indicates its (approximate) bounds

Note

Full lines indicate confidence intervals or credibility intervals (in case of a Bayesian meta-analysis). Dashed lines indicate prediction intervals. The width of all intervals is defined by the significance level chosen during meta-analysis.

References

Lewis S, Clarke M. Forest plots: trying to see the wood and the trees. BMJ. 2001; 322(7300):1479--80.
Riley RD, Higgins JPT, Deeks JJ. Interpretation of random effects meta-analyses. BMJ. 2011 342:d549--d549.

Author

Thomas Debray <thomas.debray@gmail.com>

Examples

data(Roberts)

# Frequentist random-effects meta-analysis
fit <- with(Roberts, uvmeta(r=SDM, r.se=SE, labels=rownames(Roberts)))
plot(fit)