[R-meta] Intraclass correlation
Viechtbauer Wolfgang (SP)
wolfgang.viechtbauer at maastrichtuniversity.nl
Sun Oct 1 22:55:26 CEST 2017
The sampling variances are the 'residual variances'. And since the sampling variances are heteroscedastic, there is not just one residual variance, but each study/estimate has its own sampling variance. A common approach in meta-analysis is to compute some kind of 'average/typical' sampling variance and then use that in computing various statistics (like the I^2, which is sort of an ICC). See, for example:
From: Rafael Rios [mailto:biorafaelrm at gmail.com]
Sent: Friday, 29 September, 2017 21:59
To: r-sig-meta-analysis at r-project.org; Viechtbauer Wolfgang (SP); Nelson Silva Pinto
Subject: Intraclass correlation
I'm conducting a meta-analysis involving almost 300 species. However, I had problems in insert a phylogenetic correlation matrix in the function rma.mv. Dr. Wolfgang is helping me, but I was wondering if I can calculate intraclass correlation (ICC) to evaluate the necessity to conduct a phylogenetic metanalysis. The ICC is calculated using the species variance divided by the same value plus the residual variance. When I conduct mixed models using function rma.mv, I can vizualize the variance of the random variables. However, I did not find the residual variance. Do you may clarify how may I calculate ICC using metafor package? And how may I obtain values of residual variance? Any help is welcome.
Rafael R. Moura.
Doutorando da Pós-graduação em Ecologia e Conservação de Recursos Naturais
Universidade Federal de Uberlândia, Uberlândia, MG, Brasil
Currículo Lattes: http://buscatextual.cnpq.br/buscatextual/visualizacv.do?id=K4244908A8
Research Gate: https://www.researchgate.net/profile/Rafael_Rios_Moura2
More information about the R-sig-meta-analysis