pimeta (version 1.0.1)

htsreml: Higgins-Thompson-Spiegelhalter prediction interval with a REML estimator for \(\hat{\tau}\)

Description

Higgins-Thompson-Spiegelhalter prediction interval with a REML estimator for \(\hat{\tau}\)

Usage

htsreml(y, sigma, alpha = 0.05, vartype = c("HK", "SJBC", "CL"),
  maxiter = 100)

Arguments

y

the effect size estimates vector

sigma

the within studies variances vector

alpha

the alpha level of the prediction interval

vartype

the type of the variance estimator for \(\hat{\mu}\) (default = "HK"): HK, the Hartung and Knapp (2001)'s estimator; SJBC, the Sidik and Jonkman (2006)'s bias coreccted estimator; CL, a classical estimator, \((1/\sum{\hat{w}_i})^{-1}\);

maxiter

the maximum number of iterations

Value

The average treatment effect estimate \(\hat{\mu}\) (muhat), the lower and upper prediction limits \(\hat{c}_l\) (lpi) and \(\hat{c}_u\) (upi), the REML estimator for \(\hat{\tau}\) (tau2h), and the type of the variance estimator (vartype).

References

Partlett, C, and Riley, R. D. (2017). Random effects meta-analysis: Coverage performance of 95 confidence and prediction intervals following REML estimation. Stat Med. 36(2): 301-317.

Examples

# NOT RUN {
data(sbp, package = "pimeta")
pimeta::htsreml(sbp$y, sbp$sigmak)
# $muhat
# [1] -0.3287403
# $lbpi
# [1] -0.9886995
# $ubpi
# [1] 0.3312188
# $tau2h
# [1] 0.06995102
# }