You are here

Outliers/Influential cases and Publication Bias

5 posts / 0 new
Last post
Pruealexis's picture
Offline
Joined: 08/18/2019 - 22:57
Outliers/Influential cases and Publication Bias

Hi all,
I am currently using metaSEM to create a pooled correlation matrix from 20 studies in my meta-analysis.
I previously used metafor to do this, conducting a metaanalysis via Hunter-Schmidt estimation/method on uncorrected data (maModel), and identified any outliers/influential cases. Using this method, I also completed a leave one out analysis. Metafor also enabled creating Forest Plots and Funnel Plots (to check for publication bias).
I am wondering what the alternatives are in metaSEM. I understand checking for heterogeneity, though is there a way to view my heterogeneity on a Forest plot, or to check for influential cases? How does one check for publication bias using my pooled correlation matrix? Or, do I simply go back to using metafor for this?

Thank you!

Mike Cheung's picture
Offline
Joined: 10/08/2009 - 22:37
Forest plot, and funnel plots

Forest plot, and funnel plots were developed for univariate meta-analysis. Since correlation matrices are multivariate in nature, it is not clear how these techniques can be extended to MASEM.

hannow's picture
Offline
Joined: 10/21/2020 - 03:15
Hello,

Hello,

I was wondering if there are new developments for the construction of forest and funnel plots in TSSEM? Could it be a solution to depict the respective paths (a, b, c) in individual forest and funnel plots?

Mike Cheung's picture
Offline
Joined: 10/08/2009 - 22:37
Not much yet as it is not in

Not much yet as it is not in my top priority.

You may try to plot the path coefficients or the correlations.

AdminNeale's picture
Offline
Joined: 03/01/2013 - 14:09
Plotting is easy in R

Do try to generate code yourself - there are many ways in R so all you need is code from a nice funnel plot someone else provided (google) and tweak it to taste with estimates from MetaSEM. Then you could contribute it to the project. This is how open source gets “paid” in lieu of money. Like scientific publications you can earn credit for promotions by such manifest contributions to open science. Plus you can get warm fuzzies for improving the world a bit.