MAJOR correlation (meta-analysis)

[michaelg]

Dear Kyle,
Thanks for all the great work you put into the MAJOR package! I am currently working on an meta-analysis with mutlivple variables. I was wondering if/when the multivariate meta-analyses (rma.mv() function) will become available? Thanks, Michael

[kylehamilton]

Yeah, that's in the pipeline and should be a feature by the end of the year. I need to focus on passing my candidacy review first so I might need to wait until Christmas break.

[lukas999]

Dear Kyle,

Thank you for adding the option of including categorical moderators with multiple levels. For my current meta analysis I attempt to include several moderators with 3+ levels and aim to use the major package in jamovi.

Having run into struggles with previous Jamovi versions, 1.2.16.0 seems to be working well for me but loading the major package leads to the following error message installed version is incompatible.

Edit: running older jamovi versions has not solved the issue either.

All the best,

Lukas

[jonathon]

hi lukas,

there should be the option to update to a compatible version from the jamovi library ... or are you side-loading from somewhere?

jonathon

[lukas999]

Dear Jonathon,

Thanks for the quick response. Indeed, I am sideloading major from github: https://github.com/kylehamilton/MAJOR. Loading major from the library functions but does not include the possibility of categorical moderators on correlation coefficients, which is what I am interested in.

Edit: the issue has been resolved.

[jonathon]

drop kyle a line,

you may have more success emailing him. all he needs to do is give me a commit ID, and i can update the library.

jonathon

[Jehill1]

Wondering if you could help me I'm doing a proportion meta-analysis and I have a subgroup with only two studies and when I use the DerSimonian-Laird model it says "number of parameters estimated is larger than number of observations". From my understanding you can do a proportion meta-analysis with two studies? It works okay with a fixed effects model but not with a random effects model.