How can one farming question get three different answers?
Each answer comes from a different design. A 1998–2005 trial found widespread badger culling cut cattle TB inside culled areas but raised it just outside. A difference-in-differences across 52 policy areas estimated a 56 per cent fall. A 2022 analysis by other methods found no clear effect. Ask: compared with what, who else was affected, and what exactly was done?
Last time's quick check: time zero is the moment each farm or person starts the treatment, or doesn't. Start everyone's clock at the same moment, or the comparison drifts. Bovine TB is one of the costliest problems in British cattle farming. Cattle that test positive are slaughtered, and the farm can't sell or move animals until it's clear.
Badgers can carry the infection too. So one question has divided farmers, scientists and the public for decades: does culling badgers cut TB in cattle? Here are three attempts to answer it. First, a randomised trial. From 1998 to 2005, thirty areas of English countryside were grouped in threes, and chance decided which area got widespread culling, which got culling only near outbreaks, and which got none.
Culling only near outbreaks seemed to make TB worse, and it was stopped. Widespread culling cut TB inside the culled areas, but while it went on, TB rose on farms just outside, probably because disturbed badgers roamed further. It's the spillover from film 3.5, at county scale.
From the first cull until the effect faded, cattle TB was about 29 per cent lower than in matched areas with no culling. Three years after culling stopped, the difference had gone. From 2013, culling became policy, in a staggered roll-out: areas joined year by year, where farmers applied, not where chance put them.
So a trial was no longer possible. A simple before and after won't do: in most of the worst-hit counties, TB was already falling before culling began. So in 2024, government scientists used difference-in-differences across 52 areas, comparing each area's change with areas that hadn't joined yet.
They estimated that herd TB fell by 56 per cent by the fourth year, somewhere between 41 and 69 per cent. Like every difference-in-differences, that rests on parallel trends: the assumption that, without the policy, the areas would have kept moving together. But what exactly was the treatment? From 2017, most cull areas also brought in extra blood tests for cattle, to find infected animals sooner. So this was a bundled treatment: that 56 per cent belongs to the whole policy, not to culling alone, and the authors say their analysis can't separate the parts.
A 2022 analysis of government data, using different methods, found no clear effect at all. One question, three answers. The lesson isn't to give up. It's to ask, every time: compared with what? Who else was affected? And what exactly was done? Those questions work just as well for a cover crop or a fertiliser plan. That completes the toolbox. Next, in Module 4: the machine-learning methods, in depth.
Because in most of the worst-hit counties, cattle TB was already falling before culling began, so a before-and-after would have credited the policy with a fall that was already under way. Only a comparison with areas that hadn’t yet joined could separate the two. That is what difference-in-differences does, provided the areas would otherwise have kept moving together.
Several changes made together and called one treatment. From 2017, most areas in England’s badger control policy also brought in extra blood tests for cattle, so the estimated 56 per cent fall in herd TB belongs to the whole policy, not to culling alone, and the analysis can’t separate the parts. A farm that changes tillage, cover crops and grazing in one year has the same problem.
Causal evidence is changing how farming is measured and rewarded. Leave your email and we will tell you when new films, courses and findings are published.
Prefer to read online? Vitagri also writes on Substack. See our privacy policy for how we use your email.