there's nothing wrong in principle with doing many tests at different times. It has the advantage of less effort and the disadvantage of less power (for the error variance term is bound to be inflated this way), but that problem could be overcome by increasing the sample size. Maybe I was wrong (it couldn't be told from your description) but I had the impression that you did not know what the correct sampling for the statistical analysis was.

I don't know what the correct sampling is for the statistical analysis. Would you care to help me out with that?