Opening the book…
A calibration or standard curve relates assigned concentrations to measured responses. Use appropriate standards, blank treatment, sample preparation, and a predefined model. Standards and samples must be handled comparably as required by the method, with relevant matrix and timing effects controlled.
For a straight-line model:
y = mx + b
Here, y is the measured response, x is concentration, m is slope, and b is the y-intercept. For a nonzero slope, an unknown concentration is estimated by rearranging the equation:
x = (y − b) / m
An estimate inside the supported calibration interval is interpolation. An estimate beyond that interval is extrapolation and is generally not justified for routine reporting unless the method specifically establishes its suitability. Numerical output from software does not make an out-of-range estimate valid.
An ordinary, unweighted least-squares fit to all seven CRP teaching points gives the following results. The calculation includes an intercept, uses the listed absorbances without additional blank subtraction, and is shown for instruction rather than as a validated model.
y = 0.00435188x + 0.01133083
r = 0.999826 R² = 0.999653

Calibration points, fitted line, and residuals. Residuals are measured absorbance minus predicted absorbance. The fit is a teaching calculation; it is not evidence of validated assay performance.
The correlation coefficient r measures the direction and strength of a linear association and ranges from −1 to +1. For ordinary least-squares regression with an intercept, R² describes the fraction of response variation explained by the fitted model and normally lies between 0 and 1. In simple linear regression with an intercept, R² = r². These quantities are related but are not the same statistic.
There is no universal rule that R² above 0.95 establishes an acceptable calibration. A high R² can coexist with bias, curvature, poor low-level performance, or inappropriate weighting. Inspect residuals, calibration standards, independent controls, and method-specific acceptance criteria. A residual is the observed response minus the fitted response.
For these teaching data, the high-concentration values strongly influence an unweighted fit. The fitted intercept is about 0.0113 even though the observed zero-standard response is 0.001. That difference illustrates why a high R² alone does not establish suitability across the entire concentration range.
This page titled 15.16: Standard Curves and Linear Regression was last modified on Mon, 09 Nov 2026 00:34:38 GMT and is shared under a CC BY-SA 4.0 license and was authored, remixed, and/or curated by Emalee MacKenzie.
The book could not be loaded just now.