Question 128644
I assume you use the t-distribution for confidence intervals
for population means and z-distribution for confidence 
intervals for population proportions.

Error is the standard error:
E = t*(s/sqrt(n))
Your t* depends on alpha and is always two-tailed for CI's.
Your s is the standard deviation of your sample
Your n is your sample size.
-------------------------
Cheers,
Stan H.