[R-meta] fsn with robust

[Viechtbauer, Wolfgang (NP)]

Dear Arne,

At the moment, the answer would be no. Until a month ago, there wasn't even a version of the fail-safe N method for standard random-effects models (all previous methods that have been described in the literature so far were either based on methods for pooling p-values or the equal-effects model). In the most recent update of metafor, I have actually implemented a version of fsn() that is based on a random-effects model. That in itself wasn't entirely trivial, but taking this further to multilevel/multivariate models and/or models based on robust variance estimation would be a rather big leap.

If you are interested, I'll talk about this update to fsn() (among other things) on Thursday during my upcoming live stream:

https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2023-September/004901.html

There are ways one can imagine generalizing fsn() to more complex models, but there are some conceptual and practical issues one would have to address first. If viewers on Thursday are interested, I could discuss this a bit (these live streams are always a bit dynamic, so things can go in unpredictable directions).

Best,
Wolfgang

> -----Original Message-----
> From: R-sig-meta-analysis <r-sig-meta-analysis-bounces using r-project.org> On Behalf
> Of Arne Janssen via R-sig-meta-analysis
> Sent: Tuesday, October 17, 2023 14:27
> To: r-sig-meta-analysis using r-project.org
> Cc: Arne Janssen <a.r.m.janssen using uva.nl>
> Subject: [R-meta] fsn with robust
>
> L.S.,
>
> I wonder if there is something like a fsn test for time series data that are
> analysed with a multivatiate/multilevel mixed effects model with continuous-time
> autoregression followed by a robust analysis.
>
> Best wishes,
> Arne Janssen