Linear calibration

What this model estimates

Known standards define an intercept and slope on the selected transformed axes. The calculation inverts a sample response on that line, returns it to concentration units and applies the declared sample dilution. It does not compare an unknown dose–response curve with a reference potency curve.

Use a range in which a straight relationship on the selected axes is scientifically justified. Log concentration or log response changes what “straight” means. Increasing and decreasing calibration orientations are supported.

Data and design needed

At least three distinct included standard concentrations are required, with a full-rank fitted design and usable residual variation. More replication or a wider, well-characterized range may be needed by your procedure.

Declare reportable standards, any fitting anchors, unknown/QC dilution factors and replicate treatment. The reportable endpoints come from included positive reportable standard levels. A zero standard can be a nonreportable anchor only with Untransformed concentration; a blank does not enter the fit automatically.

Check the settings

1. Select Linear calibration on Model. Check Concentration transformation, Response transformation, Weighting and Curve orientation.
2. Each axis permits Untransformed, Base-2 logarithm, Natural logarithm or Base-10 logarithm. All three weighting choices are available within their domains; weighted fitting updates expected-response weights.
3. Check the derived reportable range, any Minimum calibration R² criterion and recovery criteria. Confirm units, dilutions and the saved draft before running.

Read concentrations and intervals

Analysis summary reports the dilution-adjusted unknown and QC concentrations, intervals, reportability, suitability and contributing observations. Inspect standard back-calculation, per-well estimates, intercept, slope and residual evidence, then retrieve the PDF report.

One eligible observation uses a 95% inverse prediction interval found by inverting the Student-t individual-response prediction band. Two or more eligible observations use a Student-t log-concentration delta interval for the geometric aggregate, including shared-curve and declared response-group uncertainty. An aggregate with one eligible observation keeps that observation’s interval.

Failures and limitations

A zero or wrong-orientation slope, insufficient design, invalid transformed values or weights, nonconvergence or invalid inverse/interval can make a result unavailable. A high R² does not rule out an unsuitable calibration range or imprecise inverse prediction.

No numeric extrapolation is supplied outside the inclusive reportable interval. Inspect the range outcome and follow your procedure for a new dilution or repeat measurement rather than forcing a concentration from an out-of-range response.