Calibration point weighting

To compensate for the variance of the response at different calibration amounts, you can specify the relative weighting (or importance) of the various calibration points used to generate the curve.

The parameter that controls the weighting is Weighting Method. The default weight is equal weight for all levels and the maximum weight for each curve is normalized to 1.

The following weighting factors are available:

None

All calibration points have equal weight.

wt = 1

where
wt	Calibration level weighting factor

1/Amount

A calibration point is weighted by the factor 1/Amount, normalized to the smallest amount so that the largest weighting factor is 1. If the origin is included, it is assigned the mean of the weightings of the other calibration points.

where
Current Amount	Level amount
Minimum (Amounts)	Lowest amount across all points (levels) used for the calibration curve
wt	Calibration level weighting factor

1/Amount squared

A calibration point is weighted by the factor 1/Amount², normalized to the smallest amount so that the largest weighting factor is 1. Quadratic calibration point weightings can be used, for example, to adjust for a spread in calibration points. It makes sure that calibration points closer to the origin, which can normally be measured more accurately, get a higher weight than calibration points further away from the origin, which may be spread.

where
Current Amount	Level amount
Minimum (Amounts)	Lowest amount across all points (levels) used for the calibration curve
wt	Calibration level weighting factor

1/Response

A calibration point is weighted by the factor 1/Response, normalized to the smallest response so that the largest weighting factor is 1. If the origin is included, it is assigned the mean of the weightings of the other calibration points.

where
Current Response	Level response
Minimum (Responses)	Lowest response across all points (levels) used for the calibration curve
wt	Calibration level weighting factor

1/Response squared

A calibration point is weighted by the factor 1/Response², normalized to the smallest response so that the largest weighting factor is 1. Quadratic calibration point weightings can be used, for example, to adjust for a spread in calibration points. It makes sure that calibration points closer to the origin, which can normally be measured more accurately, get a higher weight than calibration points further away from the origin, which may be spread.

where
Current Response	Level response
Minimum (Amounts)	Lowest response across all points (levels) used for the calibration curve
wt	Calibration level weighting factor