NOEC, ECx and more - Statistical Evaluation of Biotests
Introduction to statistical methods in ecotoxicology, especially single species tests.
Aimed at
The course is aimed at all those who are involved in the evaluation of bioassays in their professional practice, either in the context of practical evaluation of data sets in the laboratory or related to the regulatory process.
The aim is to introduce the essential basic principle of statistical testing and effect threshold determination (ECx) and to enable the participants to select suitable procedures, to perform them professionally and to evaluate their results with expertise.
Concept
The course covers the statistical evaluation of single-species tests. It includes an introduction to the fundamentals of statistical testing. Special attention is given to the scaling of data, since the selection of the appropriate test depends on the underlying data scale. Basic concepts such as quantal data, metric data, test direction, error probability etc. are clarified. Building on these basics, a general scheme for selecting and performing statistical tests is presented.
Different tests for biotest evaluations are discussed. The focus is on the application in practice, i.e. concrete questions like "when which test", handling of outliers, interpretation of results.
Another main topic is the modeling of data (dose-response functions). Common models for dose-response functions (e.g., Probit, Weibull, and Logit models; multi-parameter dose-response functions) and the methods for fitting linear and nonlinear regression) will be presented and discussed, including practical hints (goodness-of-fit criteria, set points).
The theory will be illustrated and strengthened by practical exercises, and participants will work with the ToxRat Professional software. If desired, data sets from the daily practice of the course participants will be discussed.
Content
The following content can be covered in both a one-day and a two-day training course - depending on the background knowledge of the participants, on how detailed you would like certain aspects to be and on the amount of practical exercises you would like to have.
We also compile extracts and key aspects for a specific target group. No matter which aspects you need: the theory is always illustrated with practical examples.
For this purpose, we have a number of sample data sets.
If you like, concrete data sets from your practice can also be discussed.
Types of variables
Data Scales (quantal / metric data à essential for selection of statistical method)
Types of endpoints / variables in Biotests à classification metric / quantal
Basics on Statistical Testing
What is a Statistical Test?
Pairwise versus multiple testing
Error probability, test direction, alpha- and beta-error (Type I- and type II-error)
What does “significant” mean – and what does it not mean?
Statistical power – how to measure? How to affect?
Determination of a toxic threshold (NOEL/LOEL) (multiple tests)
General scheme of statistical testing (when to choose which test?)
Pretetesting: normal distribution, variance homogenity, monotonicity
Final testing – how to select an appropriate test?
Interpretation of results – important measures
Statistical Modelling – Dose/response function and point estimates (ECx)
Basics of ECx determination
Principle features of linear regression, non-linear regression
Criteria for reliability and goodness of fit
Interpretation of results
Interpolation methods
Practical demonstrations using ToxRat Professional
Discussion of own datasets of participants possible