metaBMA (version 0.3.9)

meta_bma: Model Averaging for Meta-Analysis

Description

Fits random- and fixed-effects meta-anayses and performs Bayesian model averaging for H1 (d != 0) vs. H0 (d = 0).

Usage

meta_bma(y, SE, labels = NULL, d = "norm", d.par = c(0, 0.3),
  tau = "halfcauchy", tau.par = 0.5, prior = c(1, 1, 1, 1),
  sample = 10000, summarize = "integrate", ...)

Arguments

y

mean in each study

SE

standard error in each study

labels

optional: character values with study labels

d

type of prior for mean effect \(d\) (see prior)

d.par

prior parameters for \(d\)

tau

type of prior for standard deviation of study effects \(\tau\) in random-effects meta-analysis (i.e., the SD of d across studies; see prior)

tau.par

prior parameters for \(\tau\)

prior

prior probabilities over models (possibly unnormalized) in the order c(fixed.H0, fixed.H1, random.H0, random.H1). For instance, if we expect fixed effects to be two times as likely as random effects and H0 and H1 to be equally likely: prior = c(2,2,1,1)

sample

number of samples in JAGS after burn-in and thinning (see run.jags). Samples are used to get posterior estimates for each study effect (which will show shrinkage). Only works for priors defined in prior.

summarize

whether and to compute parameter summaries (mean, median, SD, 95% quantile interval, HPD interval). If summarize = "integrate", numerical integration is used (which is precise but can require some seconds of computing time), summarize = "jags" summarizes the JAGS samples, and summarize = "none" suppresses parameter summaries.

...

arguments passed to run.jags (e.g., MCMC parameters such as sample, burnin, n.chains, thin or method="parallel")

Details

Bayesian model averaging for four meta-analysis models: Fixed- vs. random-effects and H0 (\(d=0\)) vs. H1 (e.g., \(d>0\)).

References

Gronau, Q. F., Erp, S. V., Heck, D. W., Cesario, J., Jonas, K. J., & Wagenmakers, E.-J. (2017). A Bayesian model-averaged meta-analysis of the power pose effect with informed and default priors: the case of felt power. Comprehensive Results in Social Psychology, 2(1), 123-138. https://doi.org/10.1080/23743603.2017.1326760

See Also

meta_default, meta_fixed, meta_random

Examples

# NOT RUN {
data(towels)
mb <- meta_bma(towels$logOR, towels$SE, towels$study,
               d = "norm", d.par = c(0,.3),
               tau = "halfcauchy", tau.par = .5,
               sample = 0, summarize = "none")
               # (no summary: only for CRAN checks)
mb
plot_posterior(mb, "d")
# }