The internet has determined that electronic collars are cruel. Governments are trying to ban them, veterinary organizations issue position statements, and welfare groups raise money off all the manufactured outrage. Meanwhile, the actual science, the studies these organizations cite as their justification, tells a considerably more nuanced story. A story they are not particularly interested in sharing.
I have been training dogs for decades. I have used electronic collars professionally for a significant portion of that career. I hold the only NADOI endorsement in Carroll County, Maryland, which means I have been evaluated, tested, and held accountable to a professional standard that most of the people with the loudest voices in this debate have never been subjected to.
This post is not a defense of bad training. It is not an endorsement of cranking a collar to maximum intensity and lighting a dog up like Hoover Dam. It is an intimate, honest look at what the research actually found, what it has not found, who paid for that research, and why the conclusions being drawn from it do not match the data that was revealed by these studies.
This is bound to make a lot of people grab their pitchforks and torches. I am hoping it makes them take a critical look at the lies they have been told.
What We Are Talking About When We Talk About “Shock Collars”
The term “shock collar” is the first problem. It was coined to produce a specific emotional reaction, and it works. Nobody wants to shock their dog. The word evokes cattle prods, stun guns, and deliberate cruelty. It is a politically effective label for a device that, in professional hands, functions nothing like what the label implies.
Modern electronic collars, specifically the Dogtra and E-Collar Technologies units I use in my practice, operate between 1 and 127 stimulation levels, depending on make and model. Before I fit one on a new client’s dog, I first place the receiver on the client’s palm and I tell them to let me know when they feel anything.
The average adult woman feels it around level 12. The average adult man, around 16. I do not feel it until level 20. A gymnast I evaluated one summer could not perceive any sensation until level 35.
Most dogs work between a 6 and a 12.
Between a 6 and a 12. Out of up to 127 levels of stim; below what most humans can feel at all. Well below the threshold of pain, even below the threshold of discomfort, in any meaningful clinical sense. What it delivers is a stimulus; a tap on the shoulder that says “May I have your attention, please?”
That’s not ‘shocking’. At best, it’s a mild irritant.
The distinction is not semantic either. It’s the whole argument.
The Study Behind the Bans — And Who Paid for It
When entire governments, veterinary societies, and welfare organizations cite science to justify electronic collar bans, they’re referring, almost without exception, to a body of research that traces back to a single funded project.
In 2014, Cooper, Cracknell, Hardiman, Wright, and Mills published a study in PLOS ONE titled “The Welfare Consequences and Efficacy of Training Pet Dogs with Remote Electronic Training Collars in Comparison to Reward Based Training.” This is the foundational document that started the move to ban electronic collars in the UK. It is cited by the British Veterinary Association, the Scottish Animal Welfare Commission, the RSPCA, the Kennel Club, Dogs Trust, and most of the major UK welfare organizations pressing for an English ban.
Here’s what the study’s own disclosures say about who commissioned it:
It was funded by DEFRA — the UK Department for Environment, Food and Rural Affairs — under grant AW1402, specifically described as a grant “to provide scientific evidence on which to base animal welfare policy.”
The government that wanted a ban, paid for the study that ‘justified’ the ban. That’s not independent research, it’s advocacy.
Now, let’s look at what the study actually found.
Sixty-three dogs referred for recall problems were split into three training groups. Group A was trained by industry-nominated trainers using electronic collars. Group B was trained by the same trainers without electronic collars, and Group C was trained by reward-only trainers from the Association of Pet Dog Trainers, UK.
Owner-reported success rates at the end of the study: 92% across all groups. No significant difference between e-collar and non-e-collar training in owner-reported outcomes.
Cortisol levels, both urinary and salivary, the standard physiological markers of stress, showed no significant difference between e-collar and non-e-collar groups in the controlled main study.
The headline finding that generated the policy push, that Group A dogs “spent significantly more time tense, yawned more often, and engaged in less environmental interaction,” came primarily from behavioral proxy measures, not physiological stress markers. And crucially, the sharpest welfare concerns in the study trace not to the main controlled experiment, but to a preliminary pilot phase in which high-intensity stimulation was delivered without any pre-warning cue; a use case explicitly contrary to the manufacturer’s own guidelines and training protocols.
The study that the entire UK ban campaign is built on, found no cortisol difference between groups in its properly controlled main experiment. The inflammatory findings came from testing a method no trained e-collar practitioner would use or endorse.
That’s worth reading a couple times.
The Follow-Up Study That Demolished Its Own Claims
In 2020, China, Mills, and Cooper, using the same DEFRA-funded video dataset, published a follow-up paper in Frontiers in Veterinary Science concluding that reward-based training produced “significantly better” outcomes and that “there is no evidence to indicate that e-collar training is necessary, even for its most widely cited indication.”
This paper became the single most influential document in the push to ban e-collars in Scotland and England. It was cited repeatedly in Scottish Animal Welfare Commission proceedings, in BVA lobbying materials, and in media coverage of the stalled English ban.
In 2021, researchers Rebecca Sargisson and Ian McLean at the University of Waikato published a formal peer-reviewed response in the same journal. What they found was not a disagreement about interpretation. It was a catalogue of fundamental methodological failures.
The positive-reinforcement group, the one that supposedly outperformed the e-collar group, trained at a different time, at a different location, and with different trainers than the other two groups. This is not a minor variable. It is the variable most likely to explain the difference in outcomes, and it was never disclosed in the China et al. paper.
Group assignment was only semi-random. Two dogs were moved between groups because their owners objected to the use of shock. This introduced selection bias directly into the comparison the paper’s conclusions rested on.
On Day 5 of the study, the day that drove the key statistical findings, professional trainers in the e-collar group were replaced by owners. This methodological shift was not disclosed in the China et al. paper, though it was present in the underlying data. It means the study’s critical comparison was not between training methods at all. It was between professional e-collar trainers on Day 1 through 4, owners using e-collars on Day 5, and professional reward-only trainers throughout. How is this a controlled comparison of methods? It’s a contaminating variable.
Sargisson and McLean also noted that the statistics China et al. used were inconsistent with those used by Cooper et al. in the 2014 companion paper, despite drawing from the same dataset. A Bonferroni correction for multiple comparisons, considered a standard practice, would, they argued, have likely eliminated the study’s key significant result.
Most pointedly: the dogs in the study were referred primarily for livestock-directed aggression and predatory behavior. The study measured compliance with “Sit” and “Come” commands, administered on-lead, within one meter of the trainer. That’s not the behavior the dogs were referred for. It is not the context in which the livestock attacks occured. Sargisson and McLean stated explicitly that the paper “cannot be used as an empirical justification for removing e-collars as a technique for treating dogs with behavioural problems.”
Cooper, Mills, and China published a rebuttal. They defended their statistical choices. They did not substantively address the trainer-switch variable on Day 5. They did not address the non-random group assignment. The peer-reviewed record on the most-cited study behind e-collar bans is not a vindication. It is an unresolved methodological dispute that the pro-ban community continues to treat as settled science.
What the Schalke Study Actually Found
Schalke, Stichnoth, Ott, and Jones-Baade conducted one of the most frequently cited studies in the e-collar debate, published in Applied Animal Behaviour Science. It is cited constantly by anti-e-collar advocates as evidence that electronic collars cause physiological stress.
Here is what the study found.
Fourteen Beagles were divided into three groups. Group A received stimulation contingently; paired directly with the dog’s own behavior (touching a prey decoy), making it predictable and controllable. Group H received stimulation for disobeying a recall command. Group R received stimulation randomly, with no relationship to anything the dog was doing.
Group A: no significant rise in cortisol.
Groups H and R: elevated cortisol.
The authors’ own conclusion: dogs that could “clearly associate the electric stimulus with their action” and therefore predict and control the stressor “did not show considerable or persistent stress indicators.”
This is the study anti-e-collar advocates cite as evidence that e-collars cause stress. What it actually found is that predictable, contingent, well-timed stimulation, the foundation of any professional e-collar protocol, produces no measurable physiological stress response. The elevated cortisol came from unpredictable, random stimulation: the hallmark of incompetent application, not professional use.
The Schalke study does not make the case against e-collars. It makes the case against bad training. Those aren’t the same thing and conflating them is either an error or a choice. I’m betting it was a cognitive choice.
The 2024 Study That Changed the Equation
In 2024, Johnson and Wynne, including Clive Wynne of Arizona State University, one of the leading canine cognition researchers working today, published a controlled study in Animals (MDPI) examining the efficacy of different training methods for stopping predatory chasing behavior in dogs.
Dogs with established chasing problems were randomly assigned to either an e-collar training group or one of two food-reward-only control groups.
Every dog in the e-collar group stopped chasing within one or two ten-minute training sessions. Every dog passed three of four retention and generalization tests.
Not a single dog in either food-reward-only control group successfully stopped chasing across six training sessions. Not one. Zero out of two groups.
The only observed adverse welfare indicator was a brief yelp at the moment of initial stimulation. No elevated long-term stress behaviors. No cortisol changes across groups.
This is a randomly controlled experiment, the gold standard of evidence, which shows, that for a real-world, safety-critical behavior, electronic collar training achieved results that pure positive reinforcement protocols failed entirely to achieve under matched conditions.
The response from the anti-e-collar research community was not a better study. It was commentary; a critique from Johns Hopkins questioning the protocols. An unpublished SSRN paper objecting to methodology. A statement from a Polish behaviorist association citing sample size concerns. Johnson and Wynne published a formal rebuttal in 2025, pointing out directly that the positive-reinforcement literature “is based on correlational and quasi-experimental methods that cannot determine causality.” The same methodological standard they were being held to, which the studies underlying e-collar bans conspicuously failed to meet.
There is a study, published in 2024 by a leading researcher at a major university, in a peer-reviewed journal, using random assignment, showing that e-collar training succeeded where positive reinforcement failed. The people who cite Cooper et al. (2014) and China et al. (2020) as definitive evidence against e-collars have not produced a comparable counter-study. They have only been able to snivel and whine.
England Has Not Banned E-Collars. Scotland Voted the Ban Down.
A persistent piece of misinformation circulating in training and welfare communities is that England has banned e-collars. It has not.
The Animal Welfare (Electronic Collars) (England) Regulations 2023 were approved by the House of Lords in June 2023 and were scheduled to come into force February 1, 2024. They were never brought before the House of Commons. The Statutory Instrument stalled and, as of mid-2026, has never been enacted. Electronic collar use in England remains legal under the general Animal Welfare Act 2006, subject to the standard unnecessary suffering provisions that apply to all animal welfare.
The British Veterinary Association has repeatedly pressed successive Animal Health and Welfare ministers to bring the regulations into force. IThey continue to be ignored.
In Scotland, the Scottish Parliament voted on an amendment to the Welfare of Dogs (Scotland) Bill in January 2025 that would have added an outright e-collar ban to the legislation. The amendment was voted down. Despite lobbying from the Kennel Club, the Scottish Animal Welfare Commission’s recommendation for restriction, and sustained pressure from welfare organizations, the Scottish Parliament declined to ban the tool.
The narrative that the science has settled the question and governments are simply enacting the obvious conclusion is inaccurate. England stalled. Scotland voted no. The science, as we have seen, is flawed.
The Veterinary “Consensus” Is Not a Consensus
The AVMA is the largest veterinary organization in the United States. In November 2025, AVMA President Dr. Michael Bailey stated on national radio that in extreme cases, specifically as an alternative to euthanasia for dogs with severe behavior problems, electronic collars “need to be used,” and they “need to be used in a humane fashion to make sure the [dog] understands why it’s being used.”
The American College of Veterinary Behaviorists issued a public rebuttal. They stated that the peer-reviewed literature “is clear: shock collars increase fear, stress, and aggression — and they are no more effective than reward-based training,” and that claims they can be used humanely are “not supported by science.”
The AVMA did not repudiate its president’s statement. The matter was forwarded to an internal committee for a future joint statement. As of the most recent reporting, that joint statement has not been issued.
What this exchange illustrates is not a scientific consensus being applied to policy. It is an active, unresolved, institutional dispute between major veterinary organizations about whether e-collars should ever have a legitimate professional application. The sitting president of the AVMA publicly disagreed with the ACVB’s absolutist position, in public, in late 2025. It remains an ongoing disagreement.
The International Association of Canine Professionals and the National Association of Dog Obedience Instructors maintain a formal position defending regulated, competent e-collar use, explicitly invoking Steven Lindsay’s foundational LIMA (Least Intrusive Minimally Aversive) framework and pointing out that positive-only critics have appropriated and subverted LIMA to mean something its original author explicitly did not intend. The argument is difficult to dismiss; that banning e-collars will result in the deaths of dogs whose behavior problems go unresolved because the most effective available tool has been removed from professional practice.
The Distinction Every Study Acknowledges and Every Ban Ignores
Here’s the thread that runs through every study in this literature, on every side of the debate.
Schalke et al. found no cortisol elevation when stimulation was predictable and contingent. Elevated stress appeared only with random, uncontrollable stimulation.
Salgirli, Schalke, Boehm, and Hackbarth, comparing e-collar training to prong collar and quitting-signal methods in Belgian Malinois police dogs, found that the electronic collar “induced less stress and had a stronger learning effect in comparison to the other methods” — with a critical qualifier: “in cases where the proficiency of the dog trainer is proved.” The study found, separately, that the quitting-signal method, a non-aversive approach, was “markedly stressful.” The authors’ own conclusion was that debates about training methods “should include not only the training aids but also the qualification of the trainer.” Which has been my argument all along.
China et al.’s own e-collar protocol used pre-warning cues like hand signals, vocal signals, lead pressure, and collar-borne sonic and haptic signals, before any electric stimulus was delivered. Cooper et al.’s own rebuttal described this as functioning “better as negative reinforcement” than positive punishment. The study that became the backbone of the UK ban campaign used a conservative, pre-warning, low-intensity protocol, the tap model, and still generated the conclusion that e-collars are unnecessary.
Even the studies that outright oppose e-collar use, when you read what they actually tested and what they actually found, converge on a consistent point: outcomes depend on stimulus intensity, predictability, pairing with a warning cue, and the competence of the trainer. The welfare concern isn’t inherent to the device, it’s contingent on how the device is used and by whom.
The policy responses to ban these devices, do not follow logically from the science. The science supports regulating the application, establishing practitioner standards, and distinguishing between calibrated professional use and untrained misapplication. Banning the tool removes the professional option while doing nothing about the untrained misapplication, which would likely continue with whatever aversive method is still available.
This Is Both a Tool Debate and a Literacy Debate
There’s two distinct problems in this conversation and condensing them into one is how the advocacy community sustains an argument the evidence simply can’t support.
The tool debate is real. Electronic collars, when used by a knowledgeable professional at appropriate intensity, with appropriate timing, produce behavioral outcomes that other methods cannot reliably replicate for certain high-drive, high-stakes behavior problems like predatory drift, livestock chasing, specific types of severe reactivity, and recall reliability in dogs whose intrinsic drives override standard forms of communication. Johnson and Wynne (2024) documented this under controlled conditions. Practitioners have observed it across decades of field work. The tool works. For some dogs, in some situations, it is the difference between keeping the dog and losing it.
The literacy debate is also real. An electronic collar in the hands of someone who does not understand stimulus calibration, timing, or the dog’s individual sensory threshold is not the same tool. It is a device being misused. That misuse produces the welfare outcomes the opposition points to: stress, shutdown, fear responses, and occasionally aggression. The studies that found harm were, in the majority of cases, studying precisely this; high intensity and poor timing, then generalizing these findings to professional use, which was a categorical mistake.
The answer to the literacy problem isn’t a ban. The answer is education, professional standards, and the distinction the opposition has been reluctant to make: a level 8 on a DogTra is not the same thing as a level 80. A trained practitioner is not the same as an owner who bought a $30 collar on Amazon and is playing witht the settings.
Banning the tool to address incompetent application is equivalent to banning hammers because untrained people have smashed their thumbs. The problem is the untrained person. The hammer is not the issue.
What Ecollar Use Looks Like in Practice
I have worked with dogs whose behavior problems represented genuine safety risks, to people, to livestock, to other animals, and to themselves. Dogs that had been through multiple training programs. Dogs whose owners were told, explicitly, that euthanasia was the only remaining option.
Some of those dogs are alive because of an electronic collar used at barely detectable levels where even their owners could not feel it.
The result is a dog providing the appropriate response to pressure on a level the average human cannot even register. This creates a dog whose handler can now communicate across a room, across a yard, at a distance a leash cannot cover, with a stimulus so subtle it registers as a tap on the shoulder rather than a punch in the face.
The leash is a conduit of information that flows both ways. What I teach with an electronic collar is the same thing; communication, clarity, the ability to deliver a message in a language the dog understands and respond to what the dog sends back. At perceptible levels, the collar is not a punishment device. It is an extension of the handler’s ability to communicate.
The people who have never seen this done or experienced it themselves, always have an opinion about it. That opinion is not informed by any factual observation or data.
What the Research Actually Supports
After examining the full body of literature, not the position statements derived from it, but the primary research, the evidence supports the following conclusions:
The foundational UK studies (Cooper et al. 2014; China et al. 2020) were commissioned by a government seeking to justify a policy and found no significant physiological stress difference between e-collar and non-e-collar groups in their properly controlled main experiment. The welfare findings rest primarily on behavioral proxies and, in the most dramatic cases, on a pilot phase of one study, using methods no professional practitioner endorses.
The most-cited follow-up study (China et al. 2020) has been subjected to a formal peer-reviewed methodological critique identifying non-random group assignment, an undisclosed trainer-switch diversion, and inconsistent statistics. The original authors did not substantively address the central methodological objections.
The most frequently cited stress study (Schalke et al. 2006) found no cortisol elevation under conditions of predictable, contingent, well-timed stimulation, which is precisely the conditions that define professional e-collar use.
The most recent controlled trial (Johnson and Wynne 2024) found e-collar training outperforming positive-reinforcement-only methods for predatory chasing, a finding that directly contradicts the “no evidence of necessity” claim underlying several national bans, and which was responded to with objection arguments rather than experimental replication.
The professional veterinary consensus against e-collars is not settled. The sitting president of the AVMA publicly defended humane use in late 2025, triggering an unresolved institutional dispute with the ACVB.
The ban that “happened” in England did not happen. Scotland voted the ban down in 2025.
The data, read honestly, supports a conclusion tied to intensity, timing, predictability, and trainer competence, not a conclusion that the device is inherently cruel. Multiple independent lines of evidence converge on this point. The policy and advocacy responses do not.
The Bottom Line
If you believe that any use of an electronic collar, at any level, under any conditions, by any practitioner, constitutes cruelty, you are not applying a scientific standard. You are applying a moral position dressed as one.
That position does not have the evidentiary standing it is routinely claiming to posess, in any policy debate.
The dogs in Johnson and Wynne’s study that could not be helped by positive reinforcement are not an abstraction. The dogs scheduled for euthanasia because their predatory or aggressive behavior exceeded what available tools could address are not hypothetical. The trainer-qualification variable that runs through every study in this literature repeats the consistent finding that competent, calibrated application produces materially different outcomes than incompetent application, is not a footnote. It is the central finding.
I use electronic collars. I use them at levels most humans cannot feel. I use them with precision, with timing, and with the same framework I bring to every other tool: the dog in front of me tells me what it needs, and I provide it without ideology getting in the way.
That’s not cruelty. That’s experience.
If you are in Carroll County and want to see what professional e-collar use actually looks like, reach out for a virtual consultation. If you are anywhere else in the world, virtual training sessions cover equipment selection, calibration, and the foundational mechanics of pressure-and-release communication in detail.
The conversation is worth having. Preferably with someone who has read and understands the studies, so here they are.
Sources:
Cooper et al. (2014) — PLOS ONE: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0102722
China et al. (2020) — Frontiers in Veterinary Science: https://www.frontiersin.org/journals/veterinary-science/articles/10.3389/fvets.2020.00508/full
Sargisson & McLean (2021) — Frontiers in Veterinary Science: https://www.frontiersin.org/journals/veterinary-science/articles/10.3389/fvets.2021.629746/full
Schalke et al. (2006) — Applied Animal Behaviour Science: https://happydogtraining.info/wp-content/uploads/studies/Clinical-Signs-Caused-by-the-Use-of-Electric-Training-Collars-on-Dogs-by-Schalke-et-al-2006.pdf
Salgirli et al. (2012) — Semantic Scholar: https://www.semanticscholar.org/paper/Comparison-of-learning-effects-and-stress-between-3-Salgirli-Schalke/6e2a2acbad0a3a9d1a630fa804c0e54536f17251
Johnson & Wynne (2024) — Animals (MDPI): https://pmc.ncbi.nlm.nih.gov/articles/PMC11428818/
AVMA/ACVB December 2025 dispute: https://stevedalepetworld.com/blog/avma-agrees-with-boarded-behaviorists-regarding-concerns-about-using-shock-collars/
N.A.D.O.I Position Statement https://www.nadoi.org/position-statements#training-equipment-and-methods
IACP Position Statement: https://iacpdogs.org/position-statements/
England ban status: https://joinardo.com/e-collars-are-not-banned-in-england/
Scotland vote (2025): https://www.theyworkforyou.com/sp/?id=2025-01-23.7.0
