Local regression: a new approach for measurement system comparison analysis.

Area of Science:

• Metrology
• Statistical Analysis
• Measurement Science

Background:

• Traditional analysis of measurement system comparison studies often relies on linear regression or Bland-Altman methods.
• These established approaches may be limited by the assumption of a linear relationship between measurement systems.
• There is a need for flexible analytical tools that can accommodate non-linear relationships.

Purpose of the Study:

• To propose a new graphical approach for analyzing measurement system comparison studies.
• To offer an alternative to linear regression and Bland-Altman analysis that does not assume linearity.
• To provide a method for assessing metrological equivalence between two measurement systems.

Main Methods:

• Review of established analysis approaches for measurement system comparison.
• Proposal of a difference plot incorporating non-parametric regression.
• Inclusion of pre-defined level-dependent acceptance limits for equivalence assessment.

Main Results:

• The proposed graphical approach effectively describes the relationship between two measurement systems.
• The method demonstrates potential for providing evidence of metrological equivalence.
• Illustrative data examples confirm the applicability of the approach.

Conclusions:

• The novel graphical approach offers a robust alternative for measurement system comparison.
• This method is suitable for situations where linearity cannot be assumed.
• The difference plot with non-parametric regression aids in establishing metrological equivalence.