Statistics add-in software for statistical analysis in Excel
  • Tutorials
  • Bland-Altman method comparison tutorial

Transforming the measurements to remove a relationship between differences and magnitude

When the differences are related to magnitude, it is best to try to eliminate the relationship using a transformation. Because it has a clear interpretation, logarithmic transformation is often used to remove the effect of differences increasing with magnitude. (The difference between the logarithms of two values is equivalent to the ratio of the two values.)

Other transformations, such as square root or reciprocal, cannot be clearly interpreted and are best avoided. However, rather than use a logarithmic transformation, it is usually easier to use the ratio of the measurements or the difference expressed as a percentage of the mean.

  1. On the Fit Differences task pane, in the D drop-down list, select Ratio.
  2. Click Recalculate.
    The results are recalculated and the analysis report updates.

The difference plot shows the ratio of the measurements with the points now forming a constant-width band across the measuring interval.

The mean ratio estimate is 1.10, which indicates that the Nadler method measures an average of 10% higher than the Hurley method. The limits of agreement indicate that the Nadler method may measure between ~6% to ~15% above the Hurley method for 95% of measurements.

limits of agreement ratio

Next topic: Estimating regression based limits of agreement when transformation is not enough

  •  Tutorials
  •  Distribution tutorial
  •  Correlation / PCA tutorial
  •  Compare groups means tutorial
  •  Association in 2-way contingency tables tutorial
  •  Simple linear regression tutorial
  •  Bland-Altman method comparison tutorial
  •  Plotting the relationship between methods
  •  Estimating the average bias and the limits of agreement
  •  Understanding the importance of repeatability
  •  Using repeated measurements
  •  Dealing with a relationship between difference and magnitude of measurement
  •  Transforming the measurements to remove a relationship between differences and magnitude
  •  Estimating regression based limits of agreement when transformation is not enough
  •  Estimating nonparametric limits of agreement in non-normally distributed data
  •  Estimating the precision of a measurement procedure (CLSI EP05-A3)
  •  Evaluating the linearity of a measurement procedure (CLSI EP06-A)
  •  Verifying the precision of a measurement procedure against a performance claim and estimating the bias (CLSI EP15-A3)
  •  Pareto charts tutorial
  •  Process control charts tutorial
  •  Process capability tutorial



Version 6.15
Published 18-Apr-2023
statistics software, statistical software for Excel
  • Products
  • Store 
  • Support
  • Blog
  • About us
  • Download trial
  •  Search
  •  Sign in
  •  Contact us
Analyse-it editions
  • Standard edition
  • Medical edition
  • Method Validation edition
  • Quality Control & Improvement edition
  • Ultimate edition

  • Blog  
  • About us
  • Contact us  
  • Privacy policy


Copyright 2026 Analyse-it Software, Ltd, Leeds, United Kingdom .
We use essential cookies to run the site, and optional analytics to improve the experience for visitors. For more information see our Privacy policy.