Method Comparison: Passing-Bablok & Deming Regression
Compare whether two measurement methods (e.g. a new instrument vs the gold standard) agree: Passing-Bablok (robust, no error-distribution assumption) and Deming (accounts for error on both axes) regression, giving the slope, intercept, and 95% confidence intervals — a slope 95% CI containing 1 means no proportional bias, an intercept 95% CI containing 0 means no constant bias. Computed locally in your browser; data are not uploaded.
① Enter paired data
One pair per row: method 1's value, method 2's value (space- or comma-separated). The two methods measure the same set of samples.
How to use & methodology
How do I choose between Passing-Bablok and Deming?
Both allow measurement error on both x and y (unlike ordinary least squares, which assumes error only in y). Passing-Bablok is rank-based, robust to outliers, and needs no error distribution — most common for lab method comparison; Deming assumes normal errors and needs the error-variance ratio λ. Conclusions usually agree, so report both.
How do I interpret the slope and intercept?
The slope reflects proportional bias: a 95% CI containing 1 means no proportional bias; the intercept reflects constant bias: a 95% CI containing 0 means no constant systematic error. If both hold, the two methods are broadly consistent across the measurement range and interchangeable.
How does it differ from Bland-Altman?
Regression (PB/Deming) answers 'is there systematic proportional/constant bias'; Bland-Altman answers 'how wide is the distribution of the differences and the limits of agreement'. Method validation usually combines both: regression for bias structure, BA for the range of agreement.
What value should λ (error-variance ratio) take?
Deming's λ = method 1's error variance / method 2's error variance. If the two methods are similarly precise, take 1 (this tool's default); if you know each method's replicate precision, set it from the variance ratio. λ affects the slope estimate; taking 1 when unknown is common practice.