3rd July, 2010
...In general, DoC/AHB has done little of what even they would call “research” establishing the efficacy and ecosystem safety of dropping food laced with a poison universally toxic to aerobic organisms into forest ecosystems, and what has been done is most often methodologically flawed. Little of it would probably be unpublishable except in NZ.
The focus of our investigation was on the evidence regarding the effect of aerial 1080 on native ecosystems. We have only looked superficially at the evidence regarding AHB’s claim that aerial-1080 is essential to bovine TB control, because that assertion is relatively easily dispelled (in our minds) without direct reference to the “research”, despite its dubiousness.
The evidence showing that possums are an important vector in the transmission of bovine TB is circumstantial. No direct experiments have shown transmission and the decline in infection rates in cattle have been coincident with the implementation of herd testing and control measures. Nonetheless, we have generally avoided taking on those issues because there are clear, less risky, cheaper, and equally effective means of controlling possum populations, especially in the 7 km pasture margins where it may to be important for the control of bovine TB in live stock.
Our Appraisal document is now out of date and critical new evidence and analysis has been done that greatly strengthens the case against aerial-1080. However, it does contain a section at the beginning that explains in lay terms the elements of good experimental design in complex systems.
Finally, I would warn that the principles of statistical inference and their relationship to the structure of experimental design are deep and subtle, there only being a few people in the world who truly understand them. They can be reduced to a few simple rules to be followed, but the real connection between randomization and replication in experimental design and the statistical tests is one of the greatest discoveries of the 20th Century and probably goes beyond what is achievable in a few bullet points. The reason why this is important is that it means to some degree the quality of research becomes an appeal to the authority of “experts”, which almost anyone can claim to be. (Name hidden by blogger)(******** for example is Doc’s standard reference and his is riddled with statistical and methodological errors.) Thus, pursuing the quality of research argument tends to descend into “we do good research” “no you don’t” arguments. (Incidentally, these rules are almost entirely ignored or misinterpreted in the DoC/AHB experiments.)
...In general, DoC/AHB has done little of what even they would call “research” establishing the efficacy and ecosystem safety of dropping food laced with a poison universally toxic to aerobic organisms into forest ecosystems, and what has been done is most often methodologically flawed. Little of it would probably be unpublishable except in NZ.
The focus of our investigation was on the evidence regarding the effect of aerial 1080 on native ecosystems. We have only looked superficially at the evidence regarding AHB’s claim that aerial-1080 is essential to bovine TB control, because that assertion is relatively easily dispelled (in our minds) without direct reference to the “research”, despite its dubiousness.
The evidence showing that possums are an important vector in the transmission of bovine TB is circumstantial. No direct experiments have shown transmission and the decline in infection rates in cattle have been coincident with the implementation of herd testing and control measures. Nonetheless, we have generally avoided taking on those issues because there are clear, less risky, cheaper, and equally effective means of controlling possum populations, especially in the 7 km pasture margins where it may to be important for the control of bovine TB in live stock.
Our Appraisal document is now out of date and critical new evidence and analysis has been done that greatly strengthens the case against aerial-1080. However, it does contain a section at the beginning that explains in lay terms the elements of good experimental design in complex systems.
Finally, I would warn that the principles of statistical inference and their relationship to the structure of experimental design are deep and subtle, there only being a few people in the world who truly understand them. They can be reduced to a few simple rules to be followed, but the real connection between randomization and replication in experimental design and the statistical tests is one of the greatest discoveries of the 20th Century and probably goes beyond what is achievable in a few bullet points. The reason why this is important is that it means to some degree the quality of research becomes an appeal to the authority of “experts”, which almost anyone can claim to be. (Name hidden by blogger)(******** for example is Doc’s standard reference and his is riddled with statistical and methodological errors.) Thus, pursuing the quality of research argument tends to descend into “we do good research” “no you don’t” arguments. (Incidentally, these rules are almost entirely ignored or misinterpreted in the DoC/AHB experiments.)
Dr. Quinn Whiting-Okeefe
Posts
No comments:
Post a Comment