One of the most frequent statistics quoted by DEFRA in its defence of badger culling is that ‘science’ has estimated that it would reduce the incidence of Bovine TB by around 16%.
This figure appeared suddenly, in 2011, seemingly out of the blue, was loudly proclaimed both within parliament and outside to the media, and continues to be a major plank of the pro-cull argument. DEFRA’s other major statistic, that 50% of Bovine TB is caused by badgers, was thoroughly discredited in my previous post, and to date there has been no rebuttal from the statistician involved. I therefore wanted to turn my attention to this 16% claim, and see if it was similarly based on shaky ground.
It didn’t take long! The first appearance of 16% seems to have been in a DEFRA meeting of scientific experts on 4 Apr 2011 ( for the full text use link below this post). And the minutes show that since then it has been, in that well-worn political phrase, ‘taken out of context’. Here is the relevant minute:
5. Culling conducted in line with the minimum criteria could be expected to lead to a relative reduction of confirmed new incidents of bTB in cattle herds in the local area. Even though it is not possible to give a very precise estimate, it is likely that the confirmed incidence of bTB in cattle within the culled area would be reduced relative to unculled areas by between 20-34% (see footnote 2) after 9.5 years (4 years of culling plus 5.5 years post-culling). However when taking into account the 2km perimeter area, for an idealised 150km2 area the average net benefit over 9 years would be smaller, at about 3-22%, with a central figure of 12.4% (assuming the incidence of bTB in cattle is similar inside the culled area and the 2km ring) or about 8-24% with a central figure of 16% (assuming higher incidence inside the culled area than the 2km ring) 4. Benefits would accrue over time and would be relatively small (if any) in earlier years.
In order to have a significant impact on national disease incidence, culling would need to be conducted over a very large area (bTB is currently considered endemic in over 39,000km2 of England – the area under annual bTB testing). The associated impact of culling at this scale on the national badger population is unknown.
Note the text in black bold: statistically, this 16% figure actually turns into anywhere from 3-22%, or 8-24% (the 95% confidence intervals) depending on one assumption, that there are different initial rates of bTB within and just outside a randomly-chosen cull area. To me there doesn’t seem to be any reason for assuming this, but let’s take it at face value. The ACTUAL projection, given DEFRA’s assumptions, and keeping to this 150 sq km area (the RBCT areas were 100 sq km), is more accurately stated as being anywhere between 3% and 24% (with a 95% probability).
DEFRA’s main justification for the cull, then, has been based on the higher of two averages, with, yet again (see my recent 50% post), no mention of the vagueness of this estimate as shown by the wide confidence interval.
Let me repeat: culling 70% of badgers repeated for five years in an area could statistically produce a net benefit of as low as a 3% reduction in Bovine TB, based on DEFRA’s own assumptions.
Has Owen Paterson thought to mention this 3% figure in parliament, or in media interviews? Was it even mentioned in DEFRA’s own Dec 2011 publication which was keen to emphasise the 16% figure?
Interestingly, 3% is very close to the net benefit estimated by the original ISG report based on the RBCT data. However, as yet and despite repeated requests to DEFRA, I have not been able to extract from them the statistical working which produced these projections in the first place. I have been told that are based on bTB rates noted after the RBCT had ended, and will have something to say about the way this data has been handled in due course. But meanwhile if anyone can produce a reference to the original 16% calculation I would be grateful.
27 September 2013