I intend to work on the microbiota of human breastmilk comparing two different collection methods. How do i calculate the sample size?
Depends on the question being asked and project goals. If the interest is in the most abundant microbiota only, then the sample size can be relatively small. If the interest includes microbiota that are found 1 individual in 10,000 then the sample size will need to be very large (1,000,000).
It also depends on what you are trying to do with the data. If there is also a survey asking income level, education, age, and other factors then you will need a larger sample size then if this is a descriptive study.
Except for a descriptive study, you will probably need more samples than you can collect. Look at published studies on similar topics and pay close attention to their sample size.
You can use a program like G*Power (http://gpower.hhu.de/) that will help you understand some of the issues associated with sample size if you also read the user manual. However, realize that the answers you get from this or similar programs provide a lower bound and assume that the values that you enter into the program accurately represent the values in the populations used in the new study. This is unlikely. So the answer that you get is a lower bound to a guess. It may be required by an ethics review board.
My suggestion is to look at the literature to see what is considered a good (=publishable) sample size. If possible exceed this value. It is easier to look back at your study and realize that you could have stopped a bit sooner than to look back and realize that you needed twice the number of samples.
One option that can be found in the literature is to run the study (=increase replication) until you get a significant result. This is p-hacking. Please don't do this.
Hi Sanusi,
I'd first advice that you read a few of the papers BJ Kelly or F Mattielo who have worked on this subject. The microbiome sample size estimation is not as straightforward as that of traditional biological studies.
Additionally, you can learn more and try some of the methods listed here:
http://web.cbio.uct.ac.za/~emile/AGe/Microbiome/power_MWAS.pdf
Article Power and Sample-Size Estimation for Microbiome Studies Usin...
https://fedematt.shinyapps.io/shinyMB/
Good luck,
Susheel
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size