Propagation of Error

Question

I want to determine the error (confidence interval) in the average concentration of 3 physical samples. However, I have two measures of error that I would like to combine.

1) I calculated the uncertainty in each sample concentration due to a calibration curve (prediction interval). 2) I calculated a confidence interval of the 3 physical samples.

Is there a way to combine these 2 errors?

Answer 1

Generally, absolute error grows in a "pythagorean" way: δ(A+B+...) = sqrt((δA)² + (δB)² + ...) where δ is the error estimate for that variable. I hope that helps. The absolute error of the sum is also divided by n, the total values summed.