Evaluation of correlation trend

I would like to ask you for help and advice regarding one comment I received during the peer-review process for manuscript I submitted. My colleagues and I explored the validity of new instrument for estimating habitual nutrient intake. We compared it with another (reference) method using Pearson correlation coefficient, Bland-Altman plot and cross-classification analyses.
Reviewer's question was: could the trend in bias be explored - correlation difference vs. mean nutrient intake...
We are not sure how to address this comment and we have quite short deadline...
We would appreciate any suggestion on this issue.
Thank you all in advance!


Active Member
It's not clear what the reviewer means by "could the trend in bias be explored". Does this mean that there is a discernible trend in bias and how did the reviewer know? Does it mean that if you split your intake values into a high and low group the correlations are different for the two groups? or the biases are different?
If you plot new against standard over a wide range do the points drift away from the regression line at either end? Does the scatter of points around the regression line widen as the nutrient intake increases? Either of these could indicate a change in correlation or bias.