Comment on "The structural relationship: regression in biology"

Supplementary information relevant to structural relationship regression in biology as discussed by McArdle (1988; Can. J. Zool. 66(11): 2329–2339) is presented. Although McArdle presented both linear and multiple linear regression models, we limit our comment to the linear model only. McArdle’s eq. 2 is corrected. Deming’s alternative form of the maximum likelihood (ML) solution (Deming 1943; Statistical adjustment of data. John Wiley and Sons, Inc., New York) is introduced. In the ML solution, the ratio of the mean measurement variances of the y and x variables is assumed constant over the range of the data. However, when λML is not constant, then iteratively reweighted generalized Deming regression or functional relationship estimation by maximum likelihood is more appropriate.