calcMetaPower: Calculates the statistical power of a random effects...
In MetaIntegrator: Meta-Analysis of Gene Expression Data

Description Usage Arguments Details Value Examples

Calculates the statistical power of a random effects meta-analysis based on the methods described by Valentine et al. 2010, J of Educational and Behavioral Studies.

1	calcMetaPower(es, avg_n, nStudies, hg, tail=2)

`es`	effect size you're trying to detect (e.g. 0.6)
`avg_n`	the average sample size of each GROUP in each STUDY (e.g. 10)
`nStudies`	the number of studies you put in the meta-analysis (aka Discovery cohort) (e.g. 5)
`hg`	heterogeneity, (".33" for small, "1" for moderate, & "3" for large) (e.g. 0.33)
`tail`	whether you have a one tail or two tail p-value

Based on the paper by Valentine et al.: JC Valentine, TD Pigott, and HR Rothstein. How Many Studies Do You Need? A Primer on Statistical Power for Meta-Analysis J of Educational and Behavioral Statistics April 2010 Vol 35, No 2, pp 215-247

The code itself is adapted from a blog post by Dan Quintana, Researcher at Oslo University in Biological Psychiatry On the website Towards Data Science, July 2017

https://towardsdatascience.com/how-to-calculate-statistical-power-for-your-meta-analysis-e108ee586ae8

avg_n is the average number people in each group in each study, so if you have 4 studies, and each study compared 10 cases and 10 controls, then avg_n = 10.

NOTE: THIS CODE DOES NOT TAKE MULTIPLE HYPOTHESIS TESTING INTO ACCOUNT IT ASSUMES P< 0.05

For clarity, avg_n is the average number people in each group in each study, so if you have 4 studies, and each study compared 10 cases and 10 controls, then avg_n = 10.

Statistic Power of the random effects meta-analysis described. Most statisticians want a statistical power of at least 0.8, which means that there is an 80 that if there is a true effect, you will detect it.

# effect size =0.7
# 10 samples on average in each group in each study
# 5 studies included in meta-analysis
# low heterogeneity (0.33)
calcMetaPower(es=0.7, avg_n=10, nStudies=5, hg=0.33)