What will be the ideal case for a scenario like this:

There is a control, and three experimental groups (A,B,C).

The goal of the experiment is to count how many will develop/reach a certain stage (n number of reached stage 1/ m number of starting population)

Everytime an experiment is done, the experiment could have random number of m. In addition, everytime an experiment is done, not all variables can be done.

ex. Week 1, total number of samples=20. So, only one control and experimental group a can be done with m=10 each.

Week 2, total number of samples=45. So, control, exp group A and exp group B can be done with m=15 each

The question:

Can it be that the n number of reached stage 1 for control for week 1 and week 2 be added ; and then divided for m number population for week 1 and week 2?

(n1+n2)/(m1+m2) * 100 = percentage of development

How will the error be determined for this experimental design?