next up previous contents
Next: Building the covariance matrix Up: Bypassing the Bayes' theorem Previous: Indirect measurements   Contents


Covariance matrix of experimental results

This section, based on Ref. [47], shows once more practical rules to build the covariance matrix associated with experimental data with correlated uncertainty (see also Sections [*] and [*]), treating explicitly also the case of normalization uncertainty. Then it will be shown that, in this case, the covariance matrix evaluated in this way produces biased $ \chi^2$ fits.

Subsections

Giulio D'Agostini 2003-05-15