mapower_se: Calculate statistical power using effect size and standard...

In dsquintana/metameta: A Suite of Tools for calculating and visualising study-level statistical power for meta-analyses

Calculate statistical power using effect size and standard error data

Description

Calculate statistical power using effect size and standard error data that is typically reported in meta-analyses. Power is calculated assuming the summary effect size in a meta-analysis is true effect size and assuming a range of effect sizes. The script to calculate power was adapted from the "viz_sunset" function of the metaviz package.

Usage

mapower_se(dat, size = "medium", observed_es, name)

Arguments

dat	A dataset that contains one column with observed effect sizes or outcomes labelled "yi" and a column with standard error data, labelled "sei".
size	The effect size range. The default is a medium range (0.1 to 1, in increments of 0.1). In other words, if you do not use this argument, the medium range will be used. Other options include a "small" range (0.05 to 0.5 in increments of 0.05) and a "large" range (0.25 to 2.5 in increments of 0.25).
observed_es	The observed summary effect size estimate for the meta-analysis, which is one output for the set of possible "true" effect sizes. See 'Details'
name	A label with a name for the meta-analysis, which is required for using the "firepower" function. See 'Examples'

Details

The true effect size

For the purposes of power analysis a "true" effect size needs to be specified. However, the true effect size is unknown when testing hypotheses and reported summary effect sizes are typically inflated. Here, the observed summary effect size estimate reported in the meta-analysis results is used as one possible true effect size, however, other effect sizes can be specified here instead. Additionally, statistical power for a range of true effect sizes are returned. The default is a "medium" range from 0.1 to 1, in increments of 0.1. However, a "small" and "large" range is also available (see the "size" argument below for details).

Value

This function returns the following:

dat

A dataset with results from power analyses for a range of effect sizes, including the specified observed effect size, in a column labelled "es_observed". The additional added columns include results for power analysis assuming a range of true effect sizes, with the default beginning at 0.1 ("power_es01"), then 0.2 ("power_es02"), then continuing in increments of 0.1 up to 1 ("power_es1"). A smaller or larger range of effect sizes can be specified (see the 'size' argument for details)

Examples

### An analysis using the default effect size range

power_ooi <- mapower_se(
             dat = dat_ooi,
             observed_es = 0.178,
             name = "ooi et al 2017")
power_ooi

### An analysis using the large effect size range

power_ooi_l <- mapower_se(
               dat = dat_ooi,
               size = "large",
               observed_es = 0.178,
               name = "ooi et al 2017")
power_ooi_l

### An analysis using the small effect size range

power_ooi_s <- mapower_se(
               dat = dat_ooi,
               size = "small",
               observed_es = 0.178,
               name = "ooi et al 2017")
power_ooi_s

dsquintana/metameta documentation built on Oct. 14, 2022, 2:16 a.m.