Q: What is the difference between absolute and relative weighting ?

A: When you choose Absolute Weighting, the underlying assumption is that the weights (i.e., the error affecting the data) are exactly known. If the heading of the data file is, for example:

if, on the other hand, you choose Relative Weighting, then the Fractional Standard Deviation of the data will not be taken equal to 10%. A proportionality constant will be estimated from the data themselves, and reported as the Scaled Data Variance in the Statistics Window. This type of weighting is appropriate if:

1. the error in the data is not known exactly, but its structure (e.g. constant Standard Deviation, constant Fractional Standard Deviation)  is known. Estimating the proportionality constant provides an extra degree of flexibility;
2. multiple data sets are being analyzed, and the user is uncertain about their error relative to one another.

The final FSD of the data might be larger or smaller than 10%, but it will reflect what the optimization algorithm estimates to be the error in the data, given the model.

Please see the manual and the on-line help for more details.

<< Back to Support Page