Package News for 2017
2017-07-21: R-sig-meta-analysis Mailing List
There is now a special interest group (SIG) mailing list for discussing the use of R for conducting meta-analyses at R-SIG-meta-analysis. Many questions I receive about metafor and meta-analysis in general are partly statistical, partly about programming, partly about interpreting output. Those types of questions are not well suited for the Stack Exchange sites. So, in most cases, your best option for asking question is really the R-SIG-meta-analysis mailing list.
2017-06-22: Version 2.0-0 Out
The CRAN maintainers alerted me to some issues with some of the package tests (when R is compiled with disable-long-double
). So, after addressing those issues, I pushed out the new version to CRAN (see here). The version number jump to 2.0 sounds like this is some kind of special milestone, but I don't look at it this way. I am just slowly adding new functionality and completing some of the missing pieces, one version at a time. The full changelog can be found here.
2017-04-03: Caterpillar Plots
A caterpillar plot is in essence nothing different than what is often called a forest plot in the meta-analytic literature, except that the estimates are ordered by their magnitude. You can find an example of how to draw such a plot with the metafor package here.
2017-03-19: Working on 2.0-0
I am slowly chipping away at what will become version 2.0-0 at some point in the (near?) future. Things I've already added include ranef.rma.mv()
for extracting the BLUPs of the random effects for rma.mv
models (usable at this point, but still needs a bit more work; also need to add blup.rma.mv()
), ranktest.default()
and regtest.default()
so the user can now specify the outcomes and corresponding sampling variances directly to these functions, residuals()
now has a type
argument and can compute Pearson residuals, all functions that repeatedly refit models now have the option to show a progress bar, lots of smaller tweaks/improvements, some obligatory code cleanup, and minor updates in the documentation. There will also be the option to fit what are called 'location-scale models' – introducing this feature will require a bit more explanation though, so I will defer those details to a later addition to the website.
If you already want to check out some of these features, you can always install the development version of metafor directly from GitHub (see here for instructions).