The use of statistical tools in field testing for effects of GM plants on non-target organisms (NTOs)
Data to assess whether a GM crop could have an adverse effect on non-target organisms is a mandatory part of an application for cultivation of a GM crop. Field trial data play an important role to examine the potential occurrence of such effects. If field trial data is not analysed in a statistically correct way, incorrect conclusions may be drawn. In a worst case scenario, one could incorrectly conclude that a GM crop does not pose a risk to non-target organisms.
COGEM recognises the importance of a correct statistical analysis and the difficulties one faces when trying to analyse data from field trials properly. Therefore, COGEM and the GMO office commissioned a research project on statistical methods that are suited to analyse field trial data.
The resulting research report contains important information for the assessment of environmental risks. It contains tools to determine which statistical method should be used to analyse field trial data and provides an overview of the most commonly used statistical methods. The inclusion of hyperlinks allows easy access to more detailed (background) information. Thus, amongst other things, the report contains useful background information adding to the guidelines of EFSA on statistical analysis.
COGEM points out that the research report contains information that is important when establishing
legally binding criteria for the environmental risk assessment of GM crops. The European Commission is in deliberation with its member states on a legally binding guidance that contains criteria for environmental risk assessment. The importance of the inclusion of correct statistical methods in such a guidance document is undisputed. The design of field trials has a large impact on the statistical method of choice. The information in this report might be helpful to select methods that on one hand lead to correct and statistically significant results and on the other hand do not lead to unfeasible demands with regard to field sizes and number of replicates.