Checks whether an assay reads proportionally across its measuring range by fitting polynomials to a dilution series and testing for departure from a straight line.
Over its measuring interval an assay should recover concentration linearly — double the analyte, double the reading. Linearity assessment runs a dilution/admixture series across the range and asks whether a straight line fits, or whether curvature creeps in at the ends (hook effect, saturation, matrix effects).
The engine fits first-, second-, and third-order polynomials and performs an F-test on whether the higher-order (non-linear) coefficients are jointly zero. If they are significant AND the deviation from the straight line exceeds an allowable limit at any level, the assay is non-linear over that span. It reports the deviation-from-linearity at each level so you can see where linearity breaks down.
A concentration/level column and a measured-response column, ideally with several replicates per level.
A linear/non-linear verdict, the non-linearity F-test (F, df, p), the deviation from linearity at each level, and the R² of each polynomial order.
A significant F alone is not enough — with tight replicates even a clinically trivial curvature becomes statistically significant. Judge the deviation-from-linearity against an allowable error; if the largest deviation is within your goal, treat the assay as linear for practical purposes even if F is significant.