[R-meta] Compiling different design in the same met-analysis
Viechtbauer, Wolfgang (SP)
wo||g@ng@v|echtb@uer @end|ng |rom m@@@tr|chtun|ver@|ty@n|
Tue May 4 12:14:25 CEST 2021
If one runs separate meta-analyses, one can also test for subgroup differences. This is not a distinguishing characteristic. The main difference is that separate meta-analyses automatically allow all parameters (including any variance components) to differ across analyses, while a single meta-regression model (with a categorical moderator) will by default assume that all parameters except of course for the subgroup means are the same across subgroups. But even this assumption can be relaxed and one can fit a meta-regression model that will give you exactly identical results as fitting separate meta-analyses within subgroups. See here:
The same idea generalizes to models such as those that can be fitted with rma.mv().
>From: R-sig-meta-analysis [mailto:r-sig-meta-analysis-bounces using r-project.org] On
>Behalf Of Philippe Tadger
>Sent: Tuesday, 04 May, 2021 12:04
>To: r-sig-meta-analysis using r-project.org
>Subject: Re: [R-meta] Compiling different design in the same met-analysis
>Thanks Gerta for such a simple and important reminder.
>Apart from having test for subgroup differences, which other advantage
>can have doing a subgroup analysis (with the moderator in
>meta-regression) vs separate meta-analyses?
>Just assuming that is a categorical moderator
>On 04/05/2021 11:25, Gerta Ruecker wrote:
>> Hi Gladys,
>> Note that separate meta-analyses is not the same as subgroup analysis.
>> If you do a subgroup analysis (in the Cochrane sense), you have design
>> as a moderator and obtain a treatment-design interaction test, which you
>> don't get if conducing separate analyses. Therefore I would prefer to
>> present all in one.
>> Am 04.05.2021 um 11:17 schrieb Gladys Barragan-Jason:
>>> Hi all,
>>> Thanks a lot for your responses.
>>> Actually, I did not specify it before but I am using the rma.mv
>>> <http://rma.mv> function since I can have several estimates from
>>> several studies of the same lab (random=~1|lab/study/estid).
>>> Following your recommendations, I checked whether the type of design
>>> had a significant effect on effect sizes and actually it didn't except
>>> for one specific type of intervention in which I do not have that much
>>> data: 3 papers for each design containing 7 and 4 effect sizes
>>> respectively. In this case, meta-analysis of overall estimates is
>>> non-significant while when computing them separately, one is
>>> significant (control vs. treatment groups) while the other is not
>>> (pre- vs. post treatment).
>>> I do think that would make sense to present the overall meta-analysis
>>> as well as the two designs separately ? In any case, we would need
>>> more data to conclude for sure.
More information about the R-sig-meta-analysis