I am interested in sex ratio variation. I have taken ~200 families and counted the numbers of males and females in each. Hence I have a 'population' sex ratio (one data point: the total numbers of males / total number of individuals sexed) and also a distribution of family sex ratios.
I am trying to analyse the family sex ratio distribution. This is a distribution of proportions (i.e. binary grouped data) - so should I use the σ = sqrt [ P(1 - P) / n ]? I've only ever used σ= σ / sqrt( n ) and I thought (perhaps incorrectly) that per the CLT that this latter formula could be used for any distribution with a large sample size?