Biases with the Regression Line Percentile Method and the Fallacy of a Single Standard Weight

The regression line percentile (RLP) method for computing standard weight (W(s)) equations produces third-quartile estimates that tend to be biased downward for long fish as a result of extrapolation acting in concert with systematic variation in the distribution of length-weight relationships among fish populations. Parameter mismatch in the development and use of W(s) equations (i.e., third quartile for development of W(s) equations and use of mean relative weight W(r) for application and testing) induces artifactual biases when testing equations and leads to the appearance of reduced fish condition when applied to longer fish. Consistent interpretation of mean W(r) against a single standard (e.g., third quartile, median, or any other) is not usually possible due to nonconstant variation in the distributions of mean W(r) among species and among length-classes for a given species. I recommend that future computations of W(s) use quadratic equations for all three quartiles to enable consistent and more informative interpretation of mean W(r).