var_boot_rmd: var_boot_rmd

View source: R/var_boot_rmd.R

var_boot_rmdR Documentation

var_boot_rmd

Description

Function for parametric bootstrapping procedure to estimate the variability in outcomes' effect size in case of raw mean difference as effect size measure.

Usage

var_boot_rmd(sd1i, sd2i, n1i, n2i, r, dv = 10, reps = 1000)

Arguments

sd1i

A vector of standard deviations of the outcomes in group 1 (see Details)

sd2i

A vector of standard deviations of the outcomes in group 2 (see Details)

n1i

An integer specifying the sample size of group 1

n2i

An integer specifying the sample size of group 2

r

A numerical value specifying the Pearson correlation coefficient between participants' scores on the different outcomes

dv

An integer specifying the total number of outcomes (default is 10, see Details)

reps

An integer specifying the number of bootstrap replications (default is 1,000)

Details

Multiple raw mean differences can be computed in case of two groups and multiple outcomes. The function estimates the variance of raw mean differences given a correlation among the outcomes using a parametric bootstrap procedure. For more information see van Aert & Wicherts (2023).

The vectors sd1i and sd2i can contain a single standard deviation or multiple standard deviations if information on more than one outcome is available. The integer dv is an optional argument to specify the expected number of outcomes used in a primary study. This argument can be any value between 2 and infinity. Larger values yield more accurate estimates of the variance but slow down the bootstrap procedure.

The variance that is estimated with this function can be used to correct for outcome reporting bias by including the variance as a moderator in a (multivariate) meta-analysis. Please see van Aert & Wicherts (2023) for more information.

Value

The var_boot_rmd function returns a numerical value that is an estimate of the variance of multiple correlated raw mean differences.

Author(s)

Robbie C.M. van Aert R.C.M.vanAert@tilburguniversity.edu

References

van Aert, R.C.M. & Wicherts, J.M. (2023). Correcting for outcome reporting bias in a meta-analysis: A meta-regression approach. Behavior Research Methods.

Examples

### Compute variance for an artificial example
var_boot_rmd(sd1i = c(0.8, 1.2), sd2i = c(0.85, 1.15), n1i = 100, n2i = 95, r = 0.3)


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