req_ni_r: req_ni_r

In RobbievanAert/puniform: Meta-Analysis Methods Correcting for Publication Bias

req_ni_r

Description

Function for computing the required sample size for a replication based on the Snapshot Bayesian Hybrid Meta-Analysis Method for two-independent means and raw correlation coefficients.

Usage

req_ni_r(
  ri.o,
  ni.o,
  m1i.o,
  m2i.o,
  n1i.o,
  n2i.o,
  sd1i.o,
  sd2i.o,
  tobs.o,
  alpha,
  des.pprob,
  des.pow,
  lo = 4,
  hi = 1e+05
)

Arguments

ri.o	A numeric value containing the raw correlation coefficient of the original study
ni.o	An integer containing the sample size for the raw correlation coefficient
m1i.o	A numeric value containing the mean in group 1 of the original study for two-independent means
m2i.o	A numeric value containing the mean in group 2 of the original study for two-independent means
n1i.o	A numeric value containing the sample size in group 1 of the original study for two-independent means
n2i.o	A numeric value containing the sample size in group 2 of the original study for two-independent means
sd1i.o	A numeric value containing the standard deviation in group 1 of the original study for two-independent means
sd2i.o	A numeric value containing the standard deviation in group 2 of the original study for two-independent means
tobs.o	A numeric value containing the t-value of the original study
alpha	A numeric value specifying the alpha level as used in the original study
des.pprob	A numeric value specifying the posterior probability that an user desires to obtain for one of the four true effect sizes
des.pow	A numeric value specifying the probability of observing a posterior probability larger than des.pprob that an user desires to obtain for one of the four true effect sizes
lo	A numeric value specifying the lower bound of the search interval that is used for the optimization procedure (default is 4)
hi	A numeric value specifying the upper bound of the search interval that is used for the optimization procedure (default is 100,000)

Details

The function computes the required sample size for the replication based on the Snapshot Bayesian Hybrid Meta-Analysis Method for four true effect sizes (no, small, medium, and large). The required sample size is computed by optimizing P(\pi_x \ge a)=b with \pi_x being the posterior probability with x referring to no (0), small (S), medium (M), and large (L) true effect size and a the desired posterior probability, and b the desired probability of observing a posterior probability larger than a. The required sample size for the replication is computed with and without including information of the original study. Computing the required sample size with the Snapshot Bayesian Hybrid Meta-Analysis Method is akin to computing the required sample size with a power analysis in null hypothesis significance testing. For more information see van Aert and van Assen (2016).

The req.ni.r function assumes that a two-tailed hypothesis test was conducted in the original study. In case one-tailed hypothesis tests was conducted in the original study, the alpha level has to be multiplied by two. For example, if a one-tailed hypothesis test was conducted with an alpha level of .05, an alpha of 0.1 has to be submitted to req.ni.r.

Value

The req.ni.r function returns a 4x2 matrix with in the first column the required total sample size of the replication when information of the original study is taken into account and in the second column the required sample size if information of the original study is ignored.

Author(s)

Robbie C.M. van Aert [e-mail]

References

van Aert, R.C.M. & van Assen, M.A.L.M. (2016). Bayesian evaluation of effect size after replicating an original study. Manuscript submitted for publication.

Examples

### Example as presented on page 491 in Maxwell, Lau, and Howard (2015)
req_ni_r(ri.o = 0.243, ni.o = 80, alpha = .05, des.pprob = 0.75, des.pow = 0.8)

RobbievanAert/puniform documentation built on Sept. 22, 2023, 2:53 a.m.