Sample size calculations in microbial metagenomic research

research methods design
1

In reviewing a grant for a microbial metagenomic study the applicants suggest that there is no acceptable method for undertaking this calculation: “Unfortunately, there are no generally accepted deterministic or probabilistic (simulation-based) methods for a priori calculating sample size estimates to address statistical comparisons of microbial community structures using metagenomic sequencing and massively multivariate outcomes.”

Then they later claim that due to their past experience they are certain that their study will have an “abundance” of power. Is this not a logically incongruent claim based upon the first claim? Is it true that there is no well accepted method for performing sample size calculations in metagenomic research? Shouldnt adjustments for multiple hypothesis testing be considered?

2

The “abundance of power” comment is ridiculous because power is not even the major issue. Stability of findings is. And there are many simple ways to simulate stability to determine needed sample sizes, e.g. 20  Challenges of Analyzing High-Dimensional Data – Biostatistics for Biomedical Research.

3 Likes
3

Thank you, Professor Harrell.

4

Two thoughts:

  1. That italic text reeks of LLM. But I know some actual people talk like that too.

  2. Anybody who can’t simulate their study shouldn’t be doing it. (I know this principle would rule out 98% of all medical research, but so be it.)

4 Likes