[R-meta] Subgroup analyses

Dear David

You can see answer to your question here:

https://www.metafor-project.org/doku.php/tips:models_with_or_without_intercept  
<https://www.metafor-project.org/doku.php/tips:models_with_or_without_intercept>

https://stats.stackexchange.com/questions/191144/interpreting-intercept-only-meta-analysis-model-when-effect-size-is-raw-mean  
<https://stats.stackexchange.com/questions/191144/interpreting-intercept-only-meta-analysis-model-when-effect-size-is-raw-mean>

https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2018-May/000818.html  
<https://stat.ethz.ch/pipermail/r-sig-meta-analysis/2018-May/000818.html>

I would not substract the intercept, in most of the cases is a  
strong assumption.

On 25/08/2021 12:55, David Cicchetti wrote:

Dear list members,

I am trying to run subgroup analyses in my meta analysis. I have  
the issue that I do not know whether I should include an intercept  
or not, meaning whether I should do mods = ~ Coil.Type (with  
intercept) or mods = ~ Coil.Type - 1 (without intercept).

From what I know so far we can leave out the intercept if for x = 0  
-> y = 0. Unfortunately I don't think that logic applies here. My  
variables (as for example Coil Type) do not have a 'zero'  
condition, it compares for example Coil A with Coil B.
If I want to write down the values as a table, I cannot do that  
well when including the intercept because then the estimate,  
standard error and CI are from the difference to the intercept and  
not from the level itself.

Since the Moderator tests become more easily significant when  
excluding the intercept, I was hesitant to do so. Therefore I would  
be thankful if anyone could clear me up on when/if I am allowed to  
get rid of the Intercept in my case.

Thanks in advance.

Best,
David

-- 
Kind regards/Saludos cordiales
*Philippe Tadger*
ORCID <https://orcid.org/0000-0002-1453-4105>, Reseach Gate  
<https://www.researchgate.net/profile/Philippe-Tadger>
Phone/WhatsApp: +32498774742