The Defra Code of Practice on the Spread of Ragwort has numerous problems especially from a scientific point of view. The first point to make is that it isn't the law just guidance. The somewhat oddly named Ragwort Control Act neither makes ragwort control mandatory nor makes it compulsory to make a code. It came about because of pressure from equine organisations who were, in effect, making things up about the plant. In fact even the name of the Code of Practice is wrong. It talks about the spread of a plant, implying it is increasing, when in fact it was decreasing. The Government's own Countryside Survey showed that ragwort was significantly declining during the time the name was created. This is actually what you would expect since wildflowers in general are in decline.

The most important factor in evaluating the risks is actually determining the level of the risk. In this respect the Code of Practice is very obviously wrong. It uses statistics in an invalid, and indeed innumerate fashion. This is the key important factor. The whole matter is about risk. If the risk is wrong then the whole purpose and nature of the document is undermined.

The central determination of risk is done as follows.

A figure of 500 horse deaths from ragwort poisoning in 2000. This figure is based on the number of confirmed horse deaths from ragwort poisoning seen by the Philip Leverhulme Large Animal Hospital Teaching Hospital at Liverpool University as a percentage of all the horse cases treated during the year, and grossed up to be representative of the total horse population.

This is very clearly a nonsensical use of statistics. First of all there cannot be any confirmed horse deaths from ragwort poisoning. The scientific literature is absolutely clear See: There is no test that can confirm ragwort poisoning

Secondly, Freedom of Information requests to the animal hospital concerned show that they almost never record any cases that might fit ragwort poisoning.

Thirdly, you cannot use statistics in this manner so Defra's mathematics is wrong! Examples at a hospital are not representative of the country as a whole, especially as this is the hospital of Professor Derek Knottenbelt who is well known for his work as an anti-ragwort campaigner, and who therefore would attract more cases. Well known medical expert Dr Ben Goldacre wrote an article in the Guardian Newspaper. [1] It was about human medicine, but he made a criticism of an identical idea of surveying at a medical institution and is scathing about it

We can't do a full census of the whole population every time we want some data, because they're too expensive and time-consuming. Instead, we take what we hope is a representative sample. This can fail in two interesting ways. Firstly, a sample can be systematically unrepresentative: if you want to know about the health of the population as a whole, but you survey people in a GP's waiting room, then you're an idiot.
It is perhaps the case that Defra realise their errors because on the 24th May 2016 the webmaster of gov.uk Paul Bowsher (@boffbowsh) tweeted

My task this morning: withdrawing guidance on dealing with the spread of ragwort #justGOVUKthings
[2]

 

 

 

 

 

References:

1 https://www.theguardian.com/commentisfree/2011/aug/19/bad-science-unemployment-statistical-noise

2 http://web.archive.org/web/20160524092933/https:/twitter.com/boffbowsh/status/735032168926302208