We are working on a project were the sex of sticklebacks in perch stomachs are assesed. The question here is; how do we avoid pseudoreplication? We want to find out if there is a difference in consumption of the different sexes AND we want to find out if the ratio differs from the sex-ratio found in a test fishing net. The problem is that most perch that have eaten sticklebacks have eaten more than one and thus if we use every observation (every stickleback) as a replicate we will be guilty of pseudoreplication. Chi-squared and t-tests seems like a good way of testing it then when we know that we have unbound replicates.