UKC

http://www.theguardian.com/environment/2015/mar/26/uk-drew-wrong-conclusion...

In reply to Timmd:

The government really doesn't understand the scientific process and the importance of independent peer review.

OR they understand it far too well......

In reply to Ron Rees Davies:

An ex-colleague of mine who is a biologist had a role in producing the report on the spread of TB in cattle. She suggested a number of options with the cull of badgers being a minor aside. She was apoplectic when the bulk of the report's data and conclusions were ignored and the government latched onto the badger cull as the silver bullet.
I suspect the problem is the the MPs who read these reports either don't understand them or are getting tidy incentives to reach the conclusions that benefit donors.

In reply to angry pirate:

Interests. Wearing vests, yes.

PS. This is a sad indictment of FERA:

"The study was never published in a peer-reviewed journal and has been rejected by the EU’s safety authority. Yet the Department of Environment, Food and Rural Affairs (Defra) cites it on their website as a foundation for its support of the pesticides."

I can write study proving neonicotinoids kill all known bees, dead. But if it's not subject to peer review, it's pretty questionable.

In reply to Timmd:

PS. I thought the thread was about a route: sounds like a great name...

In reply to Ron Rees Davies:

To be fair, I think that both of the studies would have gotten through peer-review and into a low-impact journal. Essentially, the original defra study was an ad-hoc attempt to make the best of a bad dataset, and either interpretation is probably a reasonable one. In both cases, they apply some relatively non-standard statistical methods to some pretty messy data and after lots of tests see a few borderline significant results but nothing that is consistent or obvious.

The one thing that both analyses (as well as the rest of the literature) show is that exposure to low-level neonic residues is a relatively unimportant determinant of bumblebee success. Either way, it isn't a result that I would be basing policy on!

In reply to Ron Rees Davies:


This has nothing to do with the government. It is down to the scientists who wrote the report, the conclusion of which will have been read by the government, or more specifically by the clerks working at the environment agency who then put forward policy proposals to the ministers.

I am also sure that party donations and lobbying by the chemical companies had no influence on the outcome of the decisions that were taken.

In reply to ByEek:

Just spotted this in the reanalysis of the defra data: https://peerj.com/articles/854/

From where he is discussing the results presented in the defra report:
"If there was no effect of pesticides then we would presumably expect the values in Table 1 to have a mean of 5 (due to type I errors where a null hypothesis is incorrectly rejected). Given that many of these values are substantially above 5, this would suggest a strong negative effect of pesticide residues on colony performance."

However the simulated values used in the original analysis were only those observations that were below the limit of detection, NOT the entire dataset. All it suggests is the same as his own analysis; that the effect is marginally significant at best. Surprised that that this got past the reviewer...

In reply to MarkJH:
I am a reasonable bright individual but I simply do not understand the meaning of what you have written. It does not surprise me that this sort of mistake was made if that is the language used to present findings to our politicians and their staff.

In reply to ByEek:

Writing policy type reports is a real skill and very different from writing papers. The wording in the original defra report is intended for both scientific and general audiences. There will certainly be civil servants with sufficient scientific expertise to understand the data as it is presented. The executive summary (intended for the non expert audience) is much more of an opinion, and will always rely to some degree on the interpretation (and integrity) of the author.

In this case, the authors were very clear that their study didn't directly answer the question it set out to address. From the final paragraph of the summary:

"This study was not a formal statistical test of the hypothesis that neonicotinoid insecticides
reduce the health of bumble bee colonies. Nevertheless, were neonicotinoids in pollen and
nectar from treated oilseed rape to be a major source of field mortality and morbidity to
bumblebee colonies, we would have expected to find a greater contribution of insecticide
residues from nearby treated crops and for there to have been a clear relationship
between observed neonicotinoid levels and measures of colony success. The absence of
these effects is reassuring but not definitive. The study underlines the importance of taking
care in extrapolating laboratory toxicology studies to the field, as well as the great need of
further studies under natural conditions.

That seems like a fairly justifiable conclusion from the data. I think that Goulson's conclusions are probably reasonable too (just with a very different spin), but he fails to address the magnitude of the effect, which is important when you are making a policy decision that will likely have significant economic and environmental consequences.

Whether you judge the UKs vote against a moratorium to be an active policy decision, or a lack of one is probably a entirely new debate.

If you are interested, the original defra report can be read here: http://fera.co.uk/ccss/documents/defraBumbleBeeReportPS2371V4a.pdf

... and Goulson's reanalysis here: https://peerj.com/articles/854/