Most dog training advice is folklore wearing a lab coat. It is passed down from television, from the dog park, from a confident trainer with a YouTube channel, and a surprising amount of it is either decades out of date or simply wrong. The frustrating part, if you are an owner trying to do right by your dog, is that some of the wrong advice still appears to work, which makes it very hard to tell good guidance from bad.
This article is the foundation for everything else on this site. It lays out what the research actually says about how dogs learn, why the popular “be the alpha” model collapsed, what controlled studies have found about reward-based and aversive methods, and what genetics and early development really contribute. Then it turns to the part almost no one talks about, which is the reason we exist as a review desk: a dog-training course is teaching two learners at once, the dog and the human, and the human is usually the harder student.
Throughout, we try to do two things at the same time: take the evidence seriously, and be fair to the many good-hearted trainers who came up through an older tradition. You can disagree with a method and still respect the person teaching it.
A dog is not running a campaign against you
Start with the single most useful idea in modern dog behavior: your dog is not plotting. When a dog pulls on the leash, ignores a recall, or counter-surfs, it is not testing your authority or trying to climb a household hierarchy. It is doing what has worked for it before. Pulling moves you forward. The counter had a sandwich on it once. Behavior is driven by its consequences, not by ambition.
This matters because the alternative model, the one that frames training as a contest for dominance, leads people toward confrontation, and confrontation backfires in measurable ways. We will get to the data. First, the mechanics.
How dogs actually learn
Two well-understood processes do almost all the work.
The first is classical conditioning, the Pavlov story everyone half-remembers. It is how a dog comes to feel about things. The leash predicts a walk, so the sight of the leash produces excitement. The vet’s office predicts discomfort, so the parking lot alone can produce fear. You are not teaching a behavior here, you are building an emotional association. This is the engine behind treating fear and reactivity, because if a trigger reliably predicts something good, the underlying feeling can change. The technical name is counterconditioning, and it is the backbone of serious behavior work.
The second is operant conditioning, which is about consequences. A behavior that pays off happens more often. A behavior that costs something happens less. Learning theorists split this into four quadrants: you can add something good (positive reinforcement), remove something unpleasant (negative reinforcement), add something unpleasant (positive punishment), or remove something good (negative punishment). All four change behavior. They do not all carry the same risks, which is the crux of the methods debate.
The practical, humane way to use operant learning is marker training, often called clicker training. The idea is simple and it solves a real problem. Dogs live in the moment, so if you reward a behavior even two seconds late, you may be rewarding whatever they are doing now instead. A marker, a click or a crisp word like “yes,” tells the dog the exact instant they got it right, and bridges the short gap until the treat arrives. Done well, it is like taking a photograph of the correct behavior. The clarity is what makes learning fast, and the mechanical skill of marking at the right moment is, not coincidentally, one of the clearest signs of whether a course is teaching you well.
If you take one thing from this section: clear, well-timed reward is not the soft option. It is the precise one.
The dominance myth, and why it refuses to die
For decades, popular training rested on the idea that dogs are pack animals locked in a struggle for rank, and that owners must establish themselves as the “alpha.” It is a tidy story. It is also built on a misreading that the original researcher spent years trying to correct.
The alpha-wolf concept came from studies of wolves housed together in captivity, where unrelated animals did form tense hierarchies. But wild wolves do not live like that. They live in families, and the “alpha” pair are simply the parents. The young follow the adults the way offspring follow parents in most species, not because they lost a fight. L. David Mech, whose early work popularized the term, has publicly and repeatedly explained that the dominance framing was a mistake. Dogs, for their part, are not wolves, and your dog is not auditing your status.
The reason the myth matters is that it produces a specific style of training built on confrontation, and confrontation has a track record. In a survey of owners of dogs with behavior problems, researchers catalogued how dogs responded to various confrontational techniques. The numbers are sobering: hitting or kicking the dog produced an aggressive response in 43 percent of cases, an “alpha roll” in 31 percent, a stare-down in 30 percent, a “dominance down” in 29 percent (Herron, Shofer and Reisner, 2009). When your training method has a one-in-three chance of provoking aggression, it is not a method, it is a risk.
The major professional bodies have moved on accordingly. The American Veterinary Society of Animal Behavior advises against dominance-based training and against referring clients to trainers who use it. If a course or trainer leans on language about being the pack leader, not letting the dog “win,” or rank reduction, they are working from a model the science discarded.
Reward versus aversion: what controlled research actually found
This is the most contested question in dog training, so we will be careful, and we will be fair.
The honest starting point is that aversive tools can work. A correction, a leash pop, a low-level shock from an electronic collar, these can change behavior, sometimes quickly. Anyone who tells you they never work is not being straight with you, and pretending otherwise just makes owners distrust the rest of the message. The reason aversives can look impressive is that avoiding discomfort is a powerful motivator, and suppressing a behavior can happen fast. “It worked for my dog” is usually a true statement.
The question that matters is not whether aversives can work. It is whether they work better, and at what cost. Here the research is consistent, and it does not favor the aversive side.
On effectiveness, a review of seventeen studies found no evidence that positive punishment is more effective than positive reinforcement, and some evidence pointing the other way (Ziv, 2017). For the specific case people argue about most, off-leash recall, a controlled trial compared dogs trained with electronic collars against dogs trained by reward-focused professionals, and concluded there was no evidence the e-collars produced better outcomes, while they did increase welfare risk (China, Mills and Cooper, 2020). If a tool that causes discomfort does not beat a tool that does not, the case for the discomfort is weak.
On cost, the evidence is clearer still. A study of ninety-two pet dogs across reward-based, mixed, and aversive-based training schools found that dogs trained with aversive methods showed more stress behaviors, spent more time in tense and low states, panted more, and had higher cortisol after training. The effects showed up outside the training context too, not just during it (Vieira de Castro et al., 2020). In plain terms, the aversive methods did not stay in the training session. They followed the dog home.
Weighing all of this, the American Veterinary Society of Animal Behavior’s 2021 position is that reward-based methods should be the first line for all training, including behavior problems, and that tools like choke, prong, and electronic collars should not be used as a first or early choice. They also note what many owners care about most: reward-based training is better for the relationship between you and your dog.
None of this requires treating a thoughtful balanced trainer as a villain. Many are skilled, many genuinely love dogs, and some teach the mechanical craft of training better than their force-free peers do. We can hold both ideas at once: respect for the teacher, and a clear-eyed view of the tools. When we review a course built around an e-collar, we will explain why it can feel effective, lay out the costs the research has documented, and point you to an approach that reaches the same goal with less risk. We go deeper on the specific tools in our guide on what the evidence says about e-collars.
What breed does, and does not, predict
People reach for breed to explain behavior, and to choose a training course. The largest study to date should make us cautious. Surveying more than eighteen thousand dogs and sequencing the DNA of more than two thousand, researchers found that breed accounts for only about 9 percent of the variation in behavior between dogs. For the traits people most want to predict, like aggression, breed was essentially useless as a predictor for an individual dog (Morrill et al., 2022).
This does not mean breed is meaningless. Tendencies exist, particularly for things like a herding dog’s interest in motion or a scent hound’s nose. But your specific dog is an individual first and a breed a distant second, and a training plan should start from the dog in front of you. When we write breed-specific guidance, we use it to correct the stereotype and refocus on the individual, because that is both more accurate and more useful.
The window that matters most
If there is one place where timing genuinely changes a dog’s life, it is early socialization. There is a sensitive period, roughly three to fourteen weeks of age, when positive exposure to people, other animals, surfaces, sounds, and gentle handling has an outsized and lasting effect on how confident and adaptable a dog becomes (the foundational work is Scott and Fuller, 1965). Miss it, and you are not doomed, but you are working uphill for a long time.
This is why veterinary behavior groups recommend starting socialization before a puppy’s vaccinations are complete, using sensible precautions and well-run puppy classes, rather than waiting. The leading cause of death for dogs under three years old is not infectious disease. It is behavior problems, through relinquishment and euthanasia. A puppy course that takes the socialization window seriously is doing the single most important job in dog training. One that treats a young puppy like a small adult to be corrected into obedience is getting the developmental science backward.
The second learner: why most online courses still fail
Here is the pivot, and it is the reason this site grades courses differently from everyone else.
Suppose a course gets all of the above right. Reward-based, humane, accurate about behavior, strong on socialization. It can still fail you completely. Not because the method is wrong, and not because your dog cannot learn, but because the course never reckoned with its second student: you.
The bottleneck in pet dog training is almost never the dog’s ability to learn. Dogs learn constantly and quickly. The bottleneck is the owner’s ability to change their own behavior, consistently, in a real living room, with a real dog, while tired and a little frustrated. That is a human learning problem, and human learning has its own well-studied science that most dog courses ignore entirely.
Consider what the research on adult learning tells us, and how routinely dog courses violate it.
Watching is not doing. The gap between knowing a technique and being able to perform it with your dog is called the transfer problem, and it is large. Information alone rarely changes behavior. Practice, feedback, and rehearsal in the real context do. A course that is ten hours of video and no practice structure has handed you a textbook and called it a driving lesson.
Attention is finite. Cognitive load theory is the simple, robust finding that working memory is small and overloads easily (Sweller). A good course sequences skills, introduces one thing at a time, and builds. A four-hour unstructured “everything about obedience” video is a cognitive-load failure no matter how good the content is, because you cannot hold it all.
Show, do not just tell. People learn complex skills better from clear demonstration with concise narration than from a talking head or a wall of text (Mayer’s work on multimedia learning). Editing and demonstration quality are not cosmetic. They are pedagogy, and they are fair to judge.
Memory needs spacing and retrieval. Skills stick through repeated, spaced practice, not a single weekend binge. Courses that build in short daily reps, repetition, and a sensible progression beat courses that explain something once and move on.
Follow-through is designed, not willed. Whether you actually do the training comes down to how easy, prompted, and routine it is (the Fogg model of behavior: motivation, ability, and a prompt). The best courses tie practice to things you already do, the morning walk, mealtimes, the ad break, so it survives a busy week. The worst rely on your motivation staying high, which it will not.
Honesty is part of the design. A course that promises a trained dog in a weekend is not just overselling, it is setting you up to fail and quit when reality arrives in week three. Realistic expectations are a feature.
This is the half of the equation the industry skips, and it is where we spend much of our attention. When we say a course is well taught, we mean it in this specific, evidence-based sense, and we explain the reasoning in our pillar on why most online courses fail.
What good training, and a good course, look like
Pulling it together, good training has a recognizable shape. It builds the behavior you want with clear, well-timed reinforcement rather than suppressing the behavior you do not want. It changes how the dog feels, not just what the dog does. It respects the animal’s emotional life, giving it agency, predictability, and a sense of safety. It meets the dog where it is developmentally. And it is honest about what it can and cannot fix.
A good course does all of that and then takes your learning as seriously as the dog’s. It sequences sensibly, shows more than it lectures, gives you something to practice and a way to tell if it is working, designs for the busy reality of your week, and tells you the truth about timelines. It is, in other words, humane and teachable at once.
These two qualities, the dog’s side and your side, are exactly the two scores we put at the top of every review. We grade each course on Method and Welfare and on Teachability and Design, because a course can be strong on one and weak on the other, and you deserve to know which. The full framework is on our methodology page, and if you would rather skip straight to a recommendation, our best online dog training courses guide puts the framework to work.
When a course is the wrong tool
A responsible guide has to say this plainly. Some problems are not suitable for a self-guided online course, and the honest move is to point you elsewhere.
Saying this costs us an affiliate sale every time, and we will keep saying it, because a site that will not tell you when not to buy is a site you cannot trust on anything else.
How we use this
Everything above is the lens for the rest of the site. When we review a course, compare two, or build a best-of list, we are applying these same ideas, scored the same way, so the pieces connect. You will see us reference this article and our topic pillars rather than re-arguing the science each time, and you will see concrete, comparative judgments, because the shared framework lets us make them. If you want to see how the scoring works in practice, start with how we evaluate courses.
The goal is not to win an argument about methods. It is to help you find the approach, and the course, that is both kind to your dog and actually teachable to you.
Selected sources: Ziv, G. (2017), Journal of Veterinary Behavior. Vieira de Castro, A.C. et al. (2020), PLOS ONE. China, L., Mills, D.S., Cooper, J.J. (2020), Frontiers in Veterinary Science. Herron, M., Shofer, F., Reisner, I. (2009), Applied Animal Behaviour Science. American Veterinary Society of Animal Behavior, Position Statement on Humane Dog Training (2021). Morrill, K. et al. (2022), Science. Scott, J.P. and Fuller, J.L. (1965), Genetics and the Social Behavior of the Dog.