Evaluating several dose-response curves with a common zero level

Abstract

Background: Experiments in different fields, including ecotoxicology, crop protection and agronomy, frequently involve qualitatively different inputs that can be applied in different doses. For each input, a separate dose-response curve can be fitted. If the dose levels include zero for each treatment, the curves must pass through the same intercept. Recently, the so-called Quenouille-Addelman solution has been proposed to analyse such data. This commentary shows a simple, previously published alternative to this method that is both easier to implement and statistically more efficient. Results: Three examples illustrate how a linear model package can be used to efficiently analyse several dose-response curves with a common intercept. Conclusion: The Quenouille-Addelman solution was originally proposed before computers when fully-fledged regression analyses posed a formidable task. That method traded convenience of computation for some loss of efficiency. With the availability of good regression packages, such a trade-off has become obsolete and a fully efficient analysis is easily obtained.