Interval estimation for proportions

Interval estimation for proportions

To estimate proportions, percentiles or rates we assume we sample from a binomial distribution with size of n with probability p of the event

Examples

XBin(n,p), find 95%CI for p

X=Yi,Y1,,YnBernoulli(p)Bin(1,p)E(X)=np,Var(X)=np(1p),n iis the sample size For a large sample size (n25), we apply CLTy¯=xn=yin=p^,Var(y^)=Var(y)n=p(1p)nZ=y^pVar(y^)=p^pp^(1p^)nN(0,1)P(plppu)=1αP(p^pup^(1p^)np^pp^(1p^)np^plp^(1p^)n)=1αP(Z1α/2p^pp^(1p^)np^plp^(1p^)n)=1αCI for p is (p^)X=Yi,Y1,,YnBernoulli(p)Bin(1,p)E(X)=np,Var(X)=np(1p),n is the sample sizeFor a large sample size (n25), we apply the Central Limit Theorem (CLT)Y¯=Xn=Yin=p^,Var(Y¯)=Var(Y)n=p(1p)nZ=p^pVar(p^)=p^pp^(1p^)nN(0,1)P(plppu)=1αP(p^pup^(1p^)np^pp^(1p^)np^plp^(1p^)n)=1αP(Z1α/2p^pp^(1p^)nZ1α/2)=1αCI for p is (p^Z1α/2p^(1p^)n,p^+Z1α/2p^(1p^)n)P(plppu)=P(p^pup(1p)nNp^pp(1p)nNp^plp(1p)nN)CI for p:(p^Z1α2p^(1p^)nN,p^+Z1α2p^(1p^)nN)