I want to know which formula can i applicate to calculate "the power analysis " of a scientific project, how many patients should be recruted (cohorts) in order to detect linkage with a significance P
A proper power analysis depends on your study design and analytic approach. It is impossible to recommend a method to calculate the sample-size needed for your study without more details. As you can see from the link Thomas provided, there are different modules for a variety of different analytic paradigms. You need to select one with a view towards what your precise study design and research questions are.
That said, you also seem to be confused over the parameters. Remember that the purpose of a "power analysis" is to determine the size of the study sample that you will require to detect a specified effect at a specified level of power.
α is not a parameter that you estimate, it is a fixed parameter. It is, in fact, the type 1 error rate of your study, or more colloquially the significance detection threshold. In your question, you specify this as "P
Thank you. But the question remains always without response. Look at the link below You will find a table that correlates between statistical power and Cohen's d. Now in my opinion the ultimate step is to do two-sample t-test to estimate the sample sizes of an experimental group and a control group (N of cohorts). So who can help me to do this test two-sample t-test?
t.test( ) function in R can do it. http://www.statmethods.net/stats/ttest.html. Are you going to do GWAS? CaTS is usually used to calculate power for gwas. http://csg.sph.umich.edu/abecasis/CaTS/tour.html. No matter which tool you use, you need to determine some parameters based on your own preliminary study or other previous studies.
Meriem: this is a different question than the one you originally posted. Please re-read my answer, because I think you are still having some trouble understanding sample size and power calculations.
Further, the table at that Wikipedia page does not show correlations of power levels and Cohen's d. It shows estimated group sizes (i.e. half the total sample size) given a t-test at a specified level of power with a specified hypothesized effect size, given by Cohen's d. In order for you to calculate your sample size, or for any of else to help you calculate it, you need to make a decision about your alpha level, power level, and the effect size you expect to see. Further, it depends on what type of t-test is appropriate for your data; what assumptions are you willing to make about the variance of your two samples?
The calculations can be performed by any statistical software or using online calculators like the one linked below. Or, in fact, for the simple case of a t-test, by hand, see link 2.