[R-meta] coef_test() and Satterthwaite correction

James Pustejovsky jepu@to @end|ng |rom gm@||@com
Mon Jan 16 18:16:50 CET 2023

Hi Sicong,

As far as I know, the Satterthwaite correction performs better than
naive-tp (or naive-t) in general, for purposes of providing hypothesis
testing procedures with close-to-nominal alpha levels and confidence
intervals with close-to-nominal coverage levels. See Pustejovsky & Tipton
(2018; https://www.jepusto.com/publication/rve-in-fixed-effects-models/)
and citations therein for simulation evidence. See also Huang & Li (2022;
https://doi.org/10.3758/s13428-021-01627-0) and Huang, Wiederman, & Zhang
(2022; https://doi.org/10.1080/00273171.2022.2077290).

Other non-Satterthwaite approaches include using saddlepoint approximation
or Edgeworth approximation. These have not been studied as extensively as
the Satterthwaite approximation, so it's less clear which should be
preferred. In my own personal experience, I have seldom seen meaningful
differences between Satterthwaite and saddlepoint approximations.


On Sun, Jan 15, 2023 at 4:03 PM Liu Sicong <64zone using gmail.com> wrote:

> Happy weekend all,
> Per reading a previous thread, I have a general question regarding
> applying the function coef_test() while selecting the method to get the
> t-test df.
>   *   I wonder if it is true that Satterthwaite correction is always the
> preferred method to get df, relative to others like naïve-tp? Would there
> be situations that one may prefer to use non-Satterthwaite approach?
> Thank you very much!
> Best regards,
> Sicong (Zone)
> ------------------------------------------
> Sicong (Zone) Liu, Ph.D.
> Research Associate, Social Action Lab<
> https://www.asc.upenn.edu/research/centers/social-action-lab>
> University of Pennsylvania
> 3620 Walnut Street,
> Philadelphia, PA 19104-6220
> ------------------------------------------
>         [[alternative HTML version deleted]]
> _______________________________________________
> R-sig-meta-analysis mailing list @ R-sig-meta-analysis using r-project.org
> To manage your subscription to this mailing list, go to:
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis

	[[alternative HTML version deleted]]

More information about the R-sig-meta-analysis mailing list