Skip to content
Prev 4984 / 5636 Next

[R-meta] Questions Re: Rma.mv Models

Hello,

Thank you all in advance for your help.  I had two minor issues with my
rma.mv models that I wanted to consult with you.

The first issue is that I am trying to plot regression plots of moderator
analyses, but the plot is missing the regression line plus the confidence
intervals that I've seen in the examples online.  Here is the code and the
graph that is generated. Do you all know why this is happening?

m_1_pub_yr <- rma.mv(yi = yi,
                     V = vi,
                     mods = ~ PUBYEAR,
                     slab = STUDY.ID,
                     data = Relationship,
                     random = ~ 1 | STUDY.ID/EFFECT_SIZE_ID,
                     test = "t",
                     method = "REML")

summary(m_1_pub_yr, digits=3)
predict(m_1_pub_yr)
metafor::regplot(m_1_pub_yr, mod="PUBYEAR", pi = TRUE, shade = TRUE,
xlim=c(1980, 2022), ylim= c(-0.75, 0.75), predlim=c(-0.75,0.75),
xlab="Publication Year", refline=0, labsize=0.9, legend=TRUE)
  title(main = "Publication Year & Parent-Child Relationship")

[image: image.png]

The second issue is that I am trying to calculate tau^2 as per requested by
the reviewers.  How does one do this?  Right now we have the following code
to determine I^2 but the reviewers want tau^2 as well:

m_3 <- rma.mv(yi = yi,
                     V = vi,
                     slab = AUTHOR,
                     data = Control,
                     random = ~ 1 | STUDY.ID/EFFECT_SIZE_ID,
                     test = "t",
                     method = "REML")

i2_3 <- var.comp(m_3)

Thank you for your help!

Best,
Tori