we are the robots
Aug29-04, 03:40 PM
I have 2 estimations (x1 and x2) of a quantity X, measured with the same instrument, with an instrumental error dx. (The same for x1 and x2).
Now I'll take the mean <x> as a representative better estimation for X.
It will have a 68% of confidence within the limits <x>+sigma and <x>-sigma, where sigma is te standard deviation. I know 2 data is too poor for this, but i have read that it's ok as long there is no better way to get an uncertainty estimation. (Is there a better one?)
Also, the interval <x> +/- (2*sigma) will have a 95% of confidence.
Of course, I'm asuming a gaussian distribution for the deviations from the true value X.
If the instrumental-induced error can be compared with the standard deviation, how can I calculate the total uncertainty?
Is it ok to get the standard deviation for an estimation of the error, or should I just calculate the error propagated in the mean formula?
Thanks.
Now I'll take the mean <x> as a representative better estimation for X.
It will have a 68% of confidence within the limits <x>+sigma and <x>-sigma, where sigma is te standard deviation. I know 2 data is too poor for this, but i have read that it's ok as long there is no better way to get an uncertainty estimation. (Is there a better one?)
Also, the interval <x> +/- (2*sigma) will have a 95% of confidence.
Of course, I'm asuming a gaussian distribution for the deviations from the true value X.
If the instrumental-induced error can be compared with the standard deviation, how can I calculate the total uncertainty?
Is it ok to get the standard deviation for an estimation of the error, or should I just calculate the error propagated in the mean formula?
Thanks.