A comparative analysis of the reduced major axis technique of fitting lines to bivariate data

Publisher Website
Google Scholar
Abstract
There are many ways of estimating the parameters of an equation to represent the relationship between two variables. While least-squares regression is generally acknowledged to be the best method to use when estimating the conditional mean of one variable given a fixed value for another, it is not usually an appropriate method to use when your primary interest is in the values of the equation parameters themselves (functional relations). In this case there are many other techniques (Barlett''s three-group method, Schnute''s trend line, the general structural relationship, major axis regression, and reduced major axis) that may provide better estimates of these values. When all of the above techniques are compared, it is found that reduced major axis is often the most applicable because of its desirable properties and ease of estimation.