I am running a 3 level analysis for my data that has clusters, but it is showing that the between cluster variance is 0%, which I believe to be in error. Anyone familiar with the 3 level meta-analysis? Also, the permutation test does not run "Error in rma.uni(x$yi, x$vi, weights = x$weights, mods = cbind(X[sample(x$k), : Fisher scoring algorithm did not converge. See 'help(rma)' for possible remedies."
This is my dataset:
Ref Test Tissue_Type Study_Type Brain_Region Pred_Out Abs_Cor Z SE n Var es.id
8 SDMT WM lesion WB static, static 0.332 0.3451 0.06 312 0.00 id_13
8 SDMT GM lesion ROI static, static 0.593 0.6823 0.06 312 0.00 id_14
9 SDMT GM lesion WB longitudinal, dynamic 0.266 0.2726 0.10 99 0.01 id_15
19 SDMT WM nonlesion ROI static, static 0.577 0.6580 0.24 20 0.06 id_33
20 SDMT GM lesion ROI static, static 0.82 1.1568 0.21 25 0.05 id_34
23 SDMT GM lesion WB longitudinal, dynamic 0.31 0.3205 0.19 31 0.04 id_38
27 SDMT GM nonlesion ROI longitudinal, dynamic 0.643 0.7633 0.32 13 0.10 id_43
35 SDMT ALL lesion WB static, static 0.557 0.6285 0.19 32 0.03 id_54
42 SDMT GM nonlesion ROI static, static 0.36 0.3769 0.14 54 0.02 id_69
42 SDMT WM nonlesion WB static, static 0.33 0.3428 0.14 54 0.02 id_70
Here is my code:
Calculate meta analysis of SDMT studies
sdmt <- subset(Meta930, Test=="SDMT")
scor=(sdmt$Abs_Cor)
sn=(sdmt$n)
s.cor <- metacor(scor,
sn,
data = sdmt,
studlab = sdmt$Ref,
sm = "COR",
method.tau = "DL")
pdf(file = "forest_scor.pdf", width = 12, height = 12)
forest.meta(s.cor,
studlab = sdmt$Ref,
comb.fixed = FALSE,
comb.random = TRUE,
print.tau2 = FALSE)
dev.off()
funnel.meta(s.cor,
comb.fixed = FALSE,
comb.random = TRUE)
eggers.test(s.cor)
Meta regression of studies (2 level)
m.qual.sdmt <- rma(yi = sdmt$Abs_Cor,
sei = sdmt$SE,
data = sdmt,
method = "ML",
mods = ~ Tissue_Type + Study_Type + Brain_Region + Pred_Out,
test = "knha")
m.qual.sdmt
permutest(m.qual.sdmt)
3 LEVEL META-ANALYSES
full.model.sdmt<- rma.mv(yi = Z,
V = Var,
slab = Ref,
data = sdmt,
random = ~ 1 | Ref/es.id,
test = "t",
method = "REML")
summary(full.model.sdmt)
convert_z2r(0.51)
r.con(0.47,10,p=.95,twotailed=TRUE)
i2.s <- var.comp(full.model.sdmt)
summary(i2.s)
plot(i2.s)
#Comparing 3 level model to 2 level model SDMT#
l3.s <- rma.mv(yi = Z,
V = Var,
slab = Ref,
data = sdmt,
random = ~ 1 | Ref/es.id,
test = "t",
method = "REML",
sigma2 = c(0, NA))
summary(l3.s)
anova(full.model.sdmt, l3.s)
And my output:
m.qual.sdmt
Mixed-Effects Model (k = 10; tau^2 estimator: ML)
tau^2 (estimated amount of residual heterogeneity): 0 (SE = 0.0031)
tau (square root of estimated tau^2 value): 0
I^2 (residual heterogeneity / unaccounted variability): 0.00%
H^2 (unaccounted variability / sampling variability): 1.00
R^2 (amount of heterogeneity accounted for): 100.00%
Test for Residual Heterogeneity:
QE(df = 4) = 2.5467, p-val = 0.6363
Test of Moderators (coefficients 2:6):
F(df1 = 5, df2 = 4) = 5.3148, p-val = 0.0653
Model Results:
estimate se tval df pval ci.lb ci.ub
intrcpt 1.0532 0.3764 2.7979 4 0.0489 0.0081 2.0983 *
Tissue_TypeGM -0.3680 0.2301 -1.5993 4 0.1850 -1.0069 0.2709
Tissue_TypeWM -0.2100 0.1544 -1.3600 4 0.2454 -0.6386 0.2187
Study_Typenonlesion -0.1243 0.0850 -1.4621 4 0.2175 -0.3603 0.1117
Brain_RegionWB -0.4026 0.1760 -2.2879 4 0.0840 -0.8911 0.0860 .
Pred_Outstatic, static -0.0936 0.1796 -0.5212 4 0.6297 -0.5921 0.4049
Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1
summary(full.model.sdmt)
Multivariate Meta-Analysis Model (k = 10; method: REML)
logLik Deviance AIC BIC AICc
-0.9462 1.8925 7.8925 8.4841 12.6925
Variance Components:
estim sqrt nlvls fixed factor
sigma^2.1 0.0000 0.0000 8 no Ref
sigma^2.2 0.0397 0.1992 10 no Ref/es.id
Test for Heterogeneity:
Q(df = 9) = 36.5793, p-val < .0001
Model Results:
estimate se tval df pval ci.lb ci.ub
0.5123 0.0800 6.4004 9 0.0001 0.3313 0.6934 ***
Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1
summary(i2.s)
% of total variance I2
Level 1 2.455693e+01 ---
Level 2 7.544307e+01 75.44
Level 3 3.806468e-08 0
Total I2: 75.44%
Any and all help appreciated. I need this to run properly for my manuscript deadline in 2 weeks. TIA.