Additional Analyses Using Example Datasets, the New Interpretive Tools, and Added Statistical Analysis

This page is an attachment to the SOP: Standard Operating Procedure for Using the NAFTA Guidance to Calculate Representative Half-life Values and for Characterizing Pesticide Degradation, Version 2.

You can find the full SOP and associated information here: Guidance to Calculate Representative Half-life Values and Characterizing Pesticide Degradation.

March 20, 2015

1. Introduction
2. New Versions of PestDF (version 0.8.13)
3. AIC and BIC Analysis
4. F Value Used for S_c

1. Introduction

   The degradation kinetics team completed analyses of example datasets. The first analyses involved running the example datasets in the new versions of PestDF to verify that the updated version would implement the new interpretive rules correctly and that the new tool was still regressing the data appropriately. The team also evaluated whether the rules used to move away from the single first-order (SFO) model were appropriate by using Akaike Information Criterion (AIC)¹ and Bayesian Information Criterion (BIC)² analysis and by testing whether changing the F value used to calculate the critical value that defines the confidence contours (S_C) would change the selected equation to describe the degradation kinetics.

   Results of all analyses are summarized in the attached Excel spreadsheet entitled Attachment 2.xlsx.

   ¹The AIC is a measure of the relative quality of a statistical model for a given set of data. It offers a relative estimate of the information lost when a given model is used to represent the process that generates the data. AIC does not test the model quality (i.e., whether a given equation provides a poor fit for a given dataset).

   ²The BIC is very similar to the AIC score. The AIC was developed based on the BIC equation.

2. New Versions of PestDF (version 0.8.13)

   Example datasets and runs using PestDF version 0.8.4, 0.8.11, and 0.8.13 are shown in the worksheet Attachment 2.xlsx. The results show that the updated interpretive rules were implemented correctly and that the tool still performs the regressions appropriately.

3. AIC and BIC Analysis

   Additional analyses of all data sets using the AIC and the BIC. Given a set of candidate models for the data, the preferred model/equation is the one with the minimum AIC or BIC value. AIC and BIC reward goodness of fit and include a penalty that is an increasing function of the number of estimated parameters. These analyses are included in the spreadsheet titled Attachment 2.xlsx for select datasets. When there was not an interpretive rule that would result in choosing a different model, or when the model was not outside of the appropriate bounds, overall AIC and BIC recommended the same model as that recommended by the NAFTA degradation kinetics.

4. F Value Used for S_c

   S_c values were calculated using a different F value (0.95 to 0.05) to determine whether changing the F value used to calculate the critical value that defines the confidence contours (S_C) would change the selected equation to describe the degradation kinetics. Results indicate that the current S_c value is appropriate and that the selected equations would not change.