# Sampling and Estimation: Question from Schweser's Notes (Q21, Reading#10)

J Foxevaluates managers who have a cross scnl popn std devn of returns of 8%. If returns are independent across managers, how large of a sample does Fox need so the std error of sample means is 1.265%? Doubt: Why is the assumption " returns are independent across managers" required?

Please don’t abbreviate so much; I have no idea what “scnl” is supposed to mean.

If the returns of one manager have strong, positive correlations with the returns of other managers, the standard error would be higher than if they were independent; if the returns have strong, negative correlations, the standard error would be lower.

Apologies. I’ll take care of that.

Is it that the formula “Standard error of sampling = population SD/ Sqroot(sample size)” is applicable only to those samples that have very less correlation among the random varibles selected in that sample? If not, how could we solve the question that I asked initially in this thread, if that assumption was not stated?

In the CFA curriculum we have no method for handling samples that aren’t independent. So, even were it not stated, you would have to assume it. I doubt that on the real exam it would ever be mentioned.

Ok. Thanks.