R Notebook

Tested whether the meta-analysis on the effect of tongkat ali on serum testosterone passed correcting for publication bias. It did not. Link to meta: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9415500/

Loading differences and standard errors

neon <- data.frame(d = c(-0.388, 1.344, 1.438, 3.783, 2.983, 0.518, 0.994, 1.352), se = c(0.313, 0.383, 0.537, 0.721, 0.678, 0.24, 0.251, 0.399))

Random effects meta-analysis

metaobj1 <- metafor::rma(yi=d, sei=se, data=neon)
summary(metaobj1)

Random-Effects Model (k = 8; tau^2 estimator: REML)

  logLik  deviance       AIC       BIC      AICc   
-11.6542   23.3084   27.3084   27.2002   30.3084   

tau^2 (estimated amount of total heterogeneity): 1.3094 (SE = 0.8080)
tau (square root of estimated tau^2 value):      1.1443
I^2 (total heterogeneity / total variability):   90.71%
H^2 (total variability / sampling variability):  10.76

Test for Heterogeneity:
Q(df = 7) = 48.6621, p-val < .0001

Model Results:

estimate      se    zval    pval   ci.lb   ci.ub     
  1.3884  0.4354  3.1891  0.0014  0.5351  2.2417  ** 

---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Funnel plot. Note the studies with larger standard errors have larger effect sizes

par(mar=c(5, 6, 4, 2))

funnel(metaobj1, xlab = "SMD", cex.lab = 1.5)

Formally testing for publication bias using the regression test

metafor::regtest(x=d, sei=se, data=neon, model='rma')

Regression Test for Funnel Plot Asymmetry

Model:     mixed-effects meta-regression model
Predictor: standard error

Test for Funnel Plot Asymmetry: z =  3.7698, p = 0.0002
Limit Estimate (as sei -> 0):   b = -1.1418 (CI: -2.4609, 0.1773)