How I can evaluate the error in sample size

Hi all, I've generated a population of NPOP rows, based on random beta distribution. Each row is normalized 1, as an array of probabilities.

This is my code for random popultation (I'm a newbie):

for i= 1:NPOP                            
        A(i)=unifrnd(0.5,2.70);           
        B(i)=unifrnd(0.5,4);              
        p = betapdf((1:7)/(7+0.1),A(i),B(i)); 
        pp= (p/sum(p));                      
        pp2= pp';
        pop(i,:)= pp2;                       
end

Now I'd like to evaluate how much error I can see if I extract randomly samples of 100,500 or 1000 rows from the population of 100,000 rows (NPOP).

Is there any measure or indicator for this comparison?

Any suggestions would be really appreciated.

Cristiano.