Statistical interpretation and graphic representation of the degradational behaviour of pesticide residues. 3. Extension of the formal description of degradation
1993
Walter, H.-F. (Bayer AG, Leverkusen (Germany). Zentrale Forschung und Entwicklung. Physik der Polymere) | Frehse, H. | Timme, G.
Two earlier publications had presented models for the mathematical description of the decline of residues, in which mathematical and graphical description of residues was achieved indirectly via empirical linearizing functions, statistical characteristics (e.g. confidence intervals) and degradation times being derived from the linear functional dependency of residues on time by calculation of regression. The present work extends the adaptation of the regression models in two directions. First it presents a linear weighted regression which allows incorporation of realistic assumptions about the analytical errors according to the level of residues: the consequences of this procedure for the various confidence intervals and degradation times (e.g. half-lives or T/10) are illustrated. It then shows how linearizing transformation could be dispensed with in the adaptation of the model to the residue values obtained from degradation studies. It also explains how to establish standard errors for the model parameters and the definition of confidence intervals, even in the event of non-linear dependency as regards the model parameters. This procedure, which does not make use of linearization, can also be applied to other non-linearizable model such as a two-zone model. The two-zone model is to be described in an subsequent publication
显示更多 [+] 显示较少 [-]