Passing-Bablok regression compares two measurement methods without assuming either is error-free — non-parametric, and robust to outliers and to error in both X and Y.
Standard OLS regression assumes the X-axis variable is measured without error and the residuals are normal — both routinely violated in method comparison, where both axes are measurements with noise and outliers are common.
Passing-Bablok estimates slope and intercept from the medians of all pairwise slopes (with sign corrections for differing signs). The result is a clean estimator that tolerates outliers up to 50% breakdown and doesn't care about distributional assumptions.
Deming regression is the parametric alternative: assumes errors in both axes are normal with a known variance ratio. Use Deming when you have prior knowledge of the relative measurement precisions; use Passing-Bablok when you don't and you want robustness.
Two numeric columns (Method A vs Method B).
Slope, intercept, bootstrap CIs for both, scatter with the fitted line and 1:1 reference.
Slope CI containing 1.0 ⇒ no proportional bias; intercept CI containing 0 ⇒ no constant bias. Both holding ⇒ the methods agree.
Report Passing-Bablok alongside the Bland-Altman plot — they answer different questions (regression vs limits of agreement).