Statistical tools for determining appropriate selection of regression models for analysis of environmental fate datasets

• Published in: Pest management science Vol. 64; no. 5; pp. 536 - 543
• Main Authors: Aldworth, Jeremy, Jackson, Scott H
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.05.2008; Wiley; Wiley Subscription Services, Inc
• Subjects: accuracy; Agronomy. Soil science and plant productions; Biological and medical sciences; data analysis; DT50; environmental degradation; environmental fate; estimation; Fundamental and applied biological sciences. Psychology; guideline study; half life; herbicides; insecticides; methodology; Models, Chemical; non-linear; nonlinear models; pesticide persistence; Pesticide Residues; Phytopathology. Animal pests. Plant and forest protection; prediction; rate constant; Regression Analysis; regression model; soil degradation; soil physics; Statistics; Studies; validity; Statistical method; guideline study; Pollutant behavior; Regression model; Environment impact; Pesticides; Regression analysis; DT50; Rate constant; Recommendation; Half life; non-linear
• ISSN: 1526-498X; 1526-4998
• DOI: 10.1002/ps.1537

BACKGROUND: In the NAFTA regulatory community, a currently consistent methodology used to estimate dissipation times for environmental fate data is not applied.RESULTS: This work demonstrates through a case study that the inappropriate use of pseudo-first-order regression models can result in inaccurate estimates of soil degradation rates, and it proposes some statistical tools that can be used to identify an appropriate statistical model to fit a particular environmental fate dataset. Diagnostic procedures have been proposed to identify the appropriate scale, and statistical testing procedures have been proposed to select the appropriate model within that scale.CONCLUSION: Results from this work demonstrate that, unless the proposed diagnostic and statistical procedures are used, inaccurate estimates of dissipation times may result. Copyright © 2008 Society of Chemical Industry