01/11/2026
I'm posting this one again as someone asked to see it and I couldn't post it to their page as it was too long.
This is a long one, but as many comparisons as possible must be taken into account to reach areasoned, balanced conclusion. I couldn't fit it onto a meme.
Did Wales’s E-Collar Ban Cause More Livestock Attacks? A Wales–England–Scotland comparison
One argument regularly made in the debate about electronic training collars is that Wales provides evidence of what happens when they are banned.
Wales prohibited the use of collars designed to administer an electric shock to dogs and cats on 24 March 2010 under the Animal Welfare (Electronic Collars) (Wales) Regulations 2010 (Welsh Government, 2010).
The claim sometimes made is straightforward: Wales banned e-collars. Livestock worrying subsequently increased. Therefore, the ban caused more livestock to be attacked or killed.
At first sight, that appears to be a simple before-and-after comparison. But there is a much better way of testing the proposition. England and Scotland did not introduce the same prohibition during this period. They therefore provide, albeit imperfectly, comparison populations.
If removing access to e-collars caused a substantial increase in livestock worrying in Wales, we ought to see a reasonably distinctive change in Wales after 2010 that was not occurring at approximately the same time in England and Scotland. That is essentially the logic of a natural experiment.
But the available evidence does not demonstrate such a pattern. So, what would we expect if the ban really caused the increase? Before looking at the figures, it is useful to define what the hypothesis predicts.
If the Welsh e-collar ban materially increased livestock worrying, we might reasonably expect to find:
1. a noticeable change in the Welsh trend following March 2010;
2. a larger or more sustained increase in Wales than in comparable areas where e-collars remained available;
3. evidence that the types of incidents increasing were incidents which e-collar training could plausibly have prevented;
4. ideally, evidence connecting offending dogs with the loss of access to electronic training.
The problem is that the available statistical evidence does not satisfy those tests.
Wales
What happened immediately around the ban?
North Wales Police provides one of the longest available police records. Their February 2026 Freedom of Information response records the following number of livestock-worrying occurrences:
Annual North Wales recorded occurrences
2006-1, 2007-2, 2008-4, 2009-5, 2010-11, 2011-16, 2012-21, 2013-12, 2014-27, 2015- 45,
2016-42, 2017- 82, 2018-110, 2019-94, 2020-73, 2021-77, 2022-100, 2023-84,2024-104,
2025-90 (North Wales Police, 2026)
As the numbers illustrate, there was clearly an increase in recorded incidents. From five occurrences in 2009, the figure increased to 11 in 2010 and 16 in 2011. But several qualifications are essential.
First, the Welsh ban did not take effect until 24 March 2010. The figure for 2010 therefore includes incidents from both before and after the ban.
Second, the increase had begun before the legislation. The recorded figures went from one in 2006 to two, four and then five before the ban was introduced. (a 500% increase!)
Third, the statistics available do not behave like a simple consequence appearing after one intervention. It rose to 21 in 2012, fell to 12 in 2013, subsequently rose substantially, fell again, and then continued to fluctuate.
Importantly, North Wales Police established the UK's first dedicated Rural Crime Team on 3 September 2013. The force subsequently became particularly active in recording, publicising and encouraging the reporting of rural crime (North Wales Police, 2023).
That date correlates far more closely with the large later increase in recorded incidents than the March 2010 e-collar legislation does.
It doesn’t prove that improved reporting explains the increase. But it means that raw police counts cannot simply be treated as a measure of the underlying prevalence of dog attacks. Unfortunately, the Welsh Government does not have a national time series which is a major problem with claims based on "Welsh statistics".
In September 2022, the Welsh Government was asked for the total number of attacks on livestock by dogs in each of the preceding five years. Its response was explicit: "The Welsh Government does not hold figures for the number of reported dog attacks on livestock." (Welsh Parliament, 2022).
So, there is no official Wales-wide dataset from which somebody can reliably produce a simple graph showing national livestock attacks before and after the 2010 ban. And whilst the North Wales Police data are valuable, North Wales is not Wales.
What about Scotland?
Scotland is particularly useful because it did not introduce a Welsh-style statutory e-collar prohibition, yet it has a much better national livestock-worrying series.
Scottish Government figures show: Yearly recorded livestock-worrying offences
2008/09-115, 2009/10- 95, 2010/11-122, 201/–12-133, 2012/13-125, 2013/14-98, 2014/15-109,
2015/16-174, 2016/17- 175, 2017/18-170, 2018/19-168, 2019/20-151, 2020/21-163
(Scottish Government, 2023; Scottish Parliament, 2020; 2021)
This comparison is very important. Between 2009–10 and 2010–11, Scotland went from 95 recorded offences to 122, an increase of approximately 28%. But Scotland had not banned e-collars.
More dramatically, recorded offences increased from 109 in 2014–15 to 174 in 2015–16- an increase of approximately 59.6% in one year. Again, there had been no Welsh-style e-collar ban.
Between 2013–14 and 2016–17 the Scottish figure increased from 98 to 175 — an increase of approximately 78.6%. Electronic collars were still legally available.
The Scottish Government cautioned that increased reporting might partly explain the change, noting that awareness of livestock worrying had increased (Scottish Government, 2023). An observation is remarkably similar to the problem we encounter in North Wales.
More recent Scottish figures tell the same story. A Scottish Government information release provides another series:
Calendar year recorded incidents (Scotland)
2020-157, 202-153, 2022-152, 2023-193, 2024- 144 (Scottish Government, 2025).
These data shouldn’t simply be joined to the earlier offence series because the reporting basis is different. But the pattern is revealing.
Scotland went: 152 → 193 → 144, representing a 27% increase followed by a fall of approximately 25%.
No corresponding introduction or removal of an e-collar ban explains those movements. Which would strongly suggest that livestock-worrying statistics plainly fluctuate for reasons other than e-collar legislation.
What about England?
England creates a different problem. Because for much of this period there is no uniform national police series.
The Office for National Statistics explained why: livestock harassment by dogs was a summary offence and police forces were not required to submit these records to the centrally held police-recorded crime database.
Consequently: "there are no uniformly recorded statistics at a national level." (Office for National Statistics, 2019).
So, claims comparing a supposed national Welsh figure with a national English figure must be treated cautiously. But we do have useful multi-force comparisons.
The National Police Chiefs’ Council looked at the series of Welsh and English forces measured together.
The NPCC examined livestock worrying in five police-force areas between approximately September 2013 and 2017: North Wales, North Yorkshire, Sussex, Devon and Cornwall, and Hertfordshire.
Across those Force areas there were:
1,705 recorded incidents; 1,928 livestock killed; 1,614 livestock injured. (NPCC, 2018).
North Wales accounted for: 449 incidents; 648 livestock killed; 376 livestock injured. (NPCC, 2018).
Subtracting North Wales leaves the four English forces with approximately:
1,256 incidents; 1,280 livestock killed;1,238 livestock injured.
So large numbers of livestock were also being attacked and killed in English police areas where e-collars remained legally available. It is not in dispute that North Wales certainly experienced a serious livestock-worrying problem. What the study does not establish is why.
It began more than three years after the Welsh e-collar ban and contains no equivalent pre-ban comparison. Consequently, it cannot demonstrate that the difference between North Wales and the English forces was significantly impacted by e-collar legislation.
There is a particularly useful 2015–17 comparison. A 2018 Farmers Weekly investigation obtained police data from forces around Britain using the same Freedom of Information exercise.
North Wales and South Wales v English Sample, recorded sheep deaths were:
2015 - 172, Wales, 106 England, 2016- 147 Wales, 74 England, 2017-135 Wales, 69 England.
The English sample consisted of Cumbria, Devon and Cornwall, Durham, Norfolk and Northamptonshire. The Welsh figures were North Wales and South Wales. (Farmers Weekly, 2018).
Now, these are not national totals and should not be presented as such. But because the figures were obtained during the same investigation, covering the same years, they offer a useful comparison.
Between 2015 and 2017: the Welsh sample fell from 172 sheep killed to 135 — a reduction of approximately 21.5%; the English sample fell from 106 to 69 — a reduction of approximately 34.9%.
So, during this directly comparable period, sheep deaths were falling on both sides of the Welsh border. Wales did not show a continuing increase while England remained protected by access to e-collars.
Who owns the dogs?
This question matters because it helps us determine which preventive measures could realistically work.
One possible explanation which is sometimes suggested for livestock worrying in Wales is that large numbers of visiting dog owners are unfamiliar with livestock and lose control of their dogs. There is surprisingly little Welsh evidence capable of quantifying this.
But the North Wales police study found something extremely important. In 89% of recorded livestock-worrying cases, the dog's owner was not present. (NPCC, 2018).
At a Welsh Government Animal Health and Welfare Framework meeting in 2018, North Wales Police representatives reported that approximately 75% of offences involved dogs escaping from inadequately secured homes or gardens, with fewer incidents involving owners deliberately allowing dogs off lead while walking (Welsh Government, 2018).
Defra subsequently reached essentially the same conclusion after discussing the Welsh situation with police and Welsh Government officials. It stated in 2023 that: "the majority of livestock worrying incidents relate to dogs escaping from their owner's properties without their knowledge." (Defra, 2023).
This has major implications for the e-collar argument. A remote-controlled collar cannot be operated by an owner who does not know that the dog has escaped.
And previous training with an e-collar should not simply be assumed to provide permanent protection against predatory behaviour in every unsupervised circumstance.
The appropriate preventive intervention for an escaping dog is first and foremost secure containment.
Scotland provides the best evidence on local owners versus visitors. The Scottish Government commissioned nationally representative research involving 1,931 sheep farmers.
Among farmers who had experienced dog attacks or chasing,
42% believed the dog belonged to a non-farming local resident.
27% believed it belonged to a visitor to the area.
7% believed it belonged to a neighbouring farm or croft.
Overall, 49% of dogs were believed to be accompanied and 38% unaccompanied, with the remainder unknown.
Most visitor-owned dogs — 80% — were believed to have been accompanied during the incident. (Scottish Government, 2019).
Visitors therefore constitute a meaningful part of the problem. But they are not the largest identified category. Locally owned dogs were, and that is consistent with the North Wales evidence showing that escaped and unaccompanied dogs constitute a major component of livestock worrying.
What does the comparison indicate?
We can now return to the original hypothesis.
1. Wales introduced an e-collar ban in March 2010. England did not. Scotland did not. Yet livestock-worrying figures subsequently rose substantially in Scotland.
2. English police forces recorded large numbers of attacks, injuries and livestock deaths while e-collars remained available.
3. During one directly comparable 2015–17 dataset, sheep deaths fell in both the Welsh and English force samples.
4. And a large proportion of North Wales incidents involved dogs whose owners were not even present.
This certainly doesn’t prove that e-collars can never be used successfully to discourage an individual dog from chasing livestock. But’ that’s a different question.
There is evidence and considerable anecdotal experience suggesting that aversive conditioning can suppress livestock chasing in some dogs.
Even the Scottish Animal Welfare Commission acknowledged some scientific and anecdotal evidence of an effect on livestock predation, while concluding that the evidence did not justify their general use and that reward-based training and environmental controls provided alternatives (Scottish Animal Welfare Commission, 2023).
But individual efficacy is not the same proposition as population-level causation.
The question here is much narrower: Did banning e-collars in Wales cause livestock worrying to increase? The available statistics do not establish that it did.
The biggest problem: reporting methods and criteria changed. North Wales illustrates why this is such a difficult question.
Its recorded occurrences went from: 1 in 2006, to 5 in 2009, to 11 in 2010, to 27 in 2014, to 110 in 2018.
It would be tempting to draw a line through those figures and declare a dramatic increase. But North Wales established Britain's first dedicated Rural Crime Team in September 2013. The force began systematically recording livestock attacks, campaigning about them, working with farmers and encouraging reporting.
The large acceleration in recorded incidents occurs after that policing development not the introduction of the e collar legislation.
Scotland experienced a similar phenomenon. Its official analysis specifically noted that increased awareness and reporting might partly explain increased recorded offences.
This is a classic epidemiological problem. An increase in detected cases is not necessarily identical to an increase in underlying incidence.
If doctors begin testing twice as many people for a disease, recorded cases can rise even if the disease itself has not become more common.
Police statistics have the same problem. More awareness, easier reporting and better recording can produce more recorded livestock attacks without an equivalent change in the true number of attacks.
Can Wales, England and Scotland really be called a natural experiment?
Only with caution. The three countries differ in geography, numbers of sheep, dog populations, tourism, access to farmland, policing structures, reporting behaviour and legislation.
England and Wales lack comparable national historical data. Police definitions have also changed.
This means we cannot conduct the ideal statistical analysis in which Wales and a matched control population show parallel pre-2010 trends before one jurisdiction suddenly introduces an e-collar ban. That evidence simply does not exist.
And that limitation cuts both ways. The data cannot demonstrate that banning e-collars reduced livestock attacks. But equally, they cannot legitimately be used to claim that banning e-collars caused attacks to increase.
What did Defra conclude?
This question was put directly to the UK Government and Defra consulted police and Welsh Government officials concerning the claimed increase in livestock worrying following the Welsh ban.
Its conclusion in July 2023 was unambiguous: "we are satisfied that there is no causal link" between the Welsh e-collar ban and livestock-worrying incidents.
Defra specifically identified two alternative explanations:
1. many incidents involved dogs escaping from owners' properties without their knowledge;
2. and police forces had increasingly encouraged the reporting of livestock-worrying incidents.
(Defra, 2023).
That conclusion fits the comparative evidence. Correlation is not causation.
There is an important distinction here. It is perfectly legitimate to say: "Recorded livestock worrying in parts of Wales increased considerably after 2010."
It is also legitimate to debate whether e-collars can sometimes suppress livestock chasing.
What is not justified by the statistics is: "Wales banned e-collars in 2010. Livestock attacks subsequently increased. Therefore, banning e-collars was significant is causing the increase."
The England and Scotland comparison makes that inference particularly difficult to sustain, because Scotland experienced substantial increases without banning e-collars.
England experienced serious livestock attacks and deaths while e-collars remained available.
Welsh livestock deaths fell during at least one directly comparable three-year period.
And many of the offending dogs in North Wales were roaming without their owners present.
In other words, the proposed cause is not necessary for the outcome to occur. The outcome occurs extensively where the supposed cause is absent.
Conclusion
Livestock worrying is a serious animal-welfare problem. Sheep and other livestock suffer injuries, stress, pregnancy loss and death. Farmers experience financial loss and considerable emotional distress.
Nothing about questioning the e-collar explanation diminishes the seriousness of those attacks. But serious subjects deserve serious evidence.
The Welsh experience does not provide convincing evidence that banning electronic shock collars caused livestock attacks to increase.
The available evidence is fragmented, reporting practices changed considerably, there is no complete Wales-wide pre- and post-ban dataset, and similar increases occurred in Scotland where e-collars remained legally available.
The strongest directly comparable evidence also shows livestock deaths falling simultaneously in samples from both England and Wales.
Perhaps most importantly, the best police evidence suggests that a large proportion of offending dogs were roaming without their owners.
The sensible conclusion is therefore not: "E-collars were banned and sheep subsequently died, therefore the ban killed sheep."
It is: "Livestock worrying is influenced by dog containment, owner behaviour, training, access to livestock, enforcement, reporting practices and probably numerous other factors. The available evidence does not isolate the Welsh e-collar ban as the cause of the observed changes."
Whilst this position is less dramatic than a meme, it is considerably closer to what the available evidence from legitimate sources demonstrates. You can see the sources of information listed below. You will note that the “Trust me bro” citation has not been included.
J J Fitzpatrick, MCGI, CFBA, BIPDT
Cosford Dog Training
15 September 2026
References
Department for Environment, Food and Rural Affairs (Defra) (2023). Written Parliamentary Question 190781, answered 3 July 2023, concerning electronic collars and livestock worrying in Wales.
Department for Environment, Food and Rural Affairs (2026). Animal Welfare Strategy for England; proposals concerning future consultation on electric shock collars.
Farmers Weekly (2018). "Special investigation: Police failing to crack down on sheep worrying", 3 April 2018.
National Police Chiefs' Council (NPCC) (2018). Livestock Worrying Police Working Group report.
North Wales Police (2023). "North Wales Police Rural Crime Team celebrate 10-year anniversary at Eisteddfod".
North Wales Police (2026). Freedom of Information response 2026/112, Livestock Worrying, providing recorded occurrences from 2006 onwards.
Office for National Statistics (2019). Freedom of Information response: Dog related harassment incidents.
Scottish Animal Welfare Commission (2023). Dog training: use of handheld remote-controlled training devices (e-collars): report. Scottish Government.
Scottish Government (2019). Sheep attacks and harassment: research. Edinburgh: Scottish Government.
Scottish Government (2025). Livestock worrying and dog attack recorded incidents: EIR release.
Scottish Parliament (2020). Written Question S5W-33611: offences recorded under the Dogs (Protection of Livestock) Act 1953.
Scottish Parliament (2021). Written Question S6W-04897: offences of protection of livestock from dogs, 2012–13 to 2020–21.
Welsh Government (2010). The Animal Welfare (Electronic Collars) (Wales) Regulations 2010, SI 2010/943 (W.97), coming into force 24 March 2010.
Welsh Government (2018). Wales Animal Health and Welfare Framework Group meeting, 14 March 2018.
Welsh Parliament (2022). Written Question WQ85944, answered 6 September 2022: reported dog attacks on livestock.
