Best way to test if method of count estimation is appropriate?

#1
So, working through some data where I have to essentially count the number of particles in each replicate. Unfortunately, sometimes there are so many particles that I need to count that it takes an unrealistic amount of time to count all of them. Because of them, I'm working through a new method where I'll estimate the total count by subsampling. I want to make sure that this subsampling method is accurate.

To do this, I'm thinking about counting a bunch of the replicates and then subsampling them using my estimation method. Then I'll calculate the CV for the full counts and the percent error for the estimation method. Would this give me a good indication of whether this method is accurate if PE is less than CV%? I'm trying to establish if the measurement error within replicates is less than the natural variability among replicates.

So in summation if:

Percent error (estimation method) < CV % (full count)

Then I'm good to go?
 
Last edited: