I would be greatful if You can give some advise regarding the following problems: As I reviewed some empirical papers, I learnt that most of them using relatively small number of observations for performing Empirical Bayes Estimation.
Question (1): Can I use it with bigger data sets?
Question (2): I am using two different types of arcsin transformations. Please find below the sources
(1):https://web.stanford.edu/~hastie/CASI_files/PDF/casi.pdf :equation 7.12, 7.13,7.21 and 7.23 (2):https://nignatiadis.github.io/assets/talks/neurips_2019_covariate_ebayes.pdf: slide 6-7
I performed the transformation and did the procedure but got different results. When I used the method from source 1 the results were negative but with source 2 positive...
source 1 main equations
Transform back:
source 2 main equations
And here We have to transform back as well.
Why I got fully different results?
Best Attila