Skip to content
Prev 510 / 5632 Next

[R-meta] standard error in predictive nonlinear meta-regression model

I would suggest you use the multivariate Delta method:

https://en.wikipedia.org/wiki/Delta_method#Multivariate_delta_method

You will have to compute some partial derivatives, which is not complicated
in your case if you have ever done it before.

Be warned though that, sometimes the Delta method approximations do not
perform well (and can underestimate the standard error). An alternative
would be to bootstrap the meta-analysis.

-Philipp


On Thu, Jan 18, 2018 at 6:07 PM, Cesar Terrer Moreno <cesar.terrer at me.com>
wrote: