Calculation of parameter uncertainties when using method of least squares

Hi everyone,

I am struggling with a maybe simple problem. I have a set of data points (x,y) for which I would like to derive a fitting function of a given type. This fitting function shall read:

y = A * ( B + C * x )^(-0.5)

with A, B, and C being the free parameters of the fitting function. I am using a quick'n'dirty approach using least squares to find the fitting function, i.e. I am running loops over A, B, and C from some minimum to maximum value for the parameters and a specified range of x. For each combination, I calculate the square error (hence least squares...) based on all data points. The combination with the least square error then is my desired fitting function.

This method works and I have processed my data successfully. Now, somebody asked me whether I could provide uncertainty ranges for A,B, and C, say I shall provide A=1.5+-0.3 etc.. But I can't imagine how to do it.

I would be glad to get some help on how to derive these uncertainties :)