question about hypothesis testing...

[semidevil]

i'm getting my concepts very confused now...

so the formula for the normal distri is (y - u)/(sigma)/((sqroot(n)).

so if I want to test y, I set that formula = to alpha(where alpha is the confidence level).

that is if it is one sided. if it is 2 sided, do I still set it to alpha, or alpha/2?

[QuantumMech]

i'm pretty sure if it's 2 sided then you set it to \frac{\alpha}{2}