Dear Researchers
How do you calculate the number of required sample size in a HRV study? If it is based on power calculation what effect size is recommended?
This depends on what type of study you want to do. if you are only describing in slguns population. By having the mean and the confidence intervals that you tolerate, you will have an N.
If, on the other hand, you are going to compare two groups. You have to look for the absolute difference between them and their standard deviation (of the difference, not of the variable). This would be the expected effect of the intervention. With this and establishing the power (beta error) and the alpha error you will be able to calculate it.
suppose I am doing the study for the first time, do you recommend looking into literature for the effect sizes? small or medium which should be preferred?
That's how it is. Whenever you can, to be more precise, it is advisable to refer to some literature. The more precise the result is where you are referring, you will probably have fewer problems finding the expected difference.
When there are many studies and with different effect sizes. I would recommend 2 options:
1- Base yourself on the best level of evidence you have: a well-done meta-analysis.
If this does not exist, 2- look for the study that has dealt with the population most similar to the one you are going to use.
When doing the sample calculation I would give more importance to this than choosing according to the magnitude of the effect.
The problems with the latter are that:
1- If you choose a very large magnitude of change, the study will be left with little power.
2- if you choose a difference that is too small, the N will be too large.
Therefore, estimating according to what was previously said, I think is better. Also ALWAYS keep common sense in mind. What really is the effect you hope to find? What difference is clinically significant? And based on that, look for the appropriate number where, mainly, you will not be left without the power to detect this clinically important difference for you.
I hope I have been useful to you, perhaps someone else can give you better advice. I believe that this statistical part also has an artisanal part of the researcher who knows the subject and uses his criteria to find the appropriate number of patients.
- Badges
- Science topic