Part of the challenge for me personally, being a vet student and passionate animal behavior geek as well as a firm believer in evidence-based medicine, is the pervasiveness of bad ideas in my field of study. Witnessing a colleague perform an "alpha roll" right in front of me, it's a struggle to balance my desire to address the issue with the need to still maintain good relationships and not become known as the token naysayer.
Dog training is one of those topics that must be handled with a delicate touch. A method isn't purely a method anymore when you're talking about its application toward an animal that a person feels a strong emotional connection with. The method becomes the person employing it, and its effectiveness becomes intrinsically tied to their value as a pet owner. Like it or not, as any trainer or behaviorist will tell you, the moment you say something like, "Dominance-based training is not as effective as we previously thought and can actually have detrimental effects on an animal" it becomes translated by the person you're talking to as, "You're a bad owner and you abuse your dog."
The problem with any topic in medicine is that bad arguments can be made to sound very persuasive and convincing by using the lingo. The argument behind dominance-based training methods is an excellent example of this (BARF diets are another good example). Advocates such as Cesar Millan point to wolf pack hierarchy models as an example of "natural" applications of dominance-based behavioral conditioning. They tell dog owners to be their dog's "alpha" by using techniques employed by wolves such as throat holds and alpha rolls. They also attempt to shame owners by telling them that disobedience is a form of dominance which proves that their dog doesn't respect their status as "pack leader." The appeal to nature fallacy is something we skeptics are well aware of but it is unfortunately remarkably persuasive with the general public.
A huge, glaring problem with the dominance hierarchy argument is that it makes the assumption that behavior models which we have obtained based on the study of captive wolf packs are reflective of natural behavior in the wild. This is patently false. Firstly, the dominance-based hierarchy suggested by Millan only occurs in captive wolf packs. Wolf packs in the wild consist of genetically related members with the breeding pair being the "alphas." The frequent displays of aggression and dominance seen in captivity do not occur in a natural setting. Secondly, feral dog "packs" - the aggregates formed by stray dogs - do not display this hierarchy model, so even if it were true of wolves in the wild this model does not appear applicable for domestic canines. (Mech, 1999; Taylor & Francis, 2004)
And then there's the problem with the word "dominance" itself. Common usage would lead most people to believe that dominance is a personality trait; something a dog just is. A common thing we hear from our clients is, "She's just so dominant!" Or claim that their dog is trying to be dominant over them. Dominance has a very specific meaning within the context of animal behavior and it isn't something an animal just is. This is a common misunderstanding and something I've even seen my colleagues use. Dr. Sophia Yin, a DVM with a Master's in animal behavior and a widely renowned expert in dog behavior does a pretty good job of summing it up here. She has written extensively on the topics of dominance, aggression and training and I highly encourage anyone with a dog to spend several hours reading her articles. The American Veterinary Society of Animal Behavior reinforces Dr. Yin's position with their official statement on dominance theory:
“Dominance is defined as a relationship between individual animals that is established by force/aggression and submission, to determine who has priority access to multiple resources such as food, preferred resting spots, and mates (Bernstein 1981; Drews 1993)... In our relationship with our pets, priority access to resources is not the major concern. The majority of behaviors owners want to modify, such as excessive vocalization, unruly greetings, and failure to come when called, are not related to valued resources and may not even involve aggression. Rather, these behaviors occur because they have been inadvertently rewarded and because alternate appropriate behaviors have not been trained instead.”
But beyond the implausibility of the theory behind the use of dominance and physically aversive stimuli in dog training, as well as the misuse of the term "dominance", there is the added factor that it just doesn't have a wide range of practical use. Meaning in the majority of cases, it doesn't work. Several recent studies have confirmed that dominance/positive punishment training methods have a number of negative effects on dogs (including physical injury and death in cases of choke chains and prong collars being used incorrectly) and can actually impair learning ability. These methods also cause fear and escalate aggression in terms of frequency, magnitude and situational aggression - meaning a dog that wasn't previously aggressive becomes aggressive, or a conditionally aggressive dog begins to display aggression in situations where it previously did not (Husson et al, 2009; Hiby et al, 2004; AVSAB, 2008). This is particularly worrisome for vets and shelter workers. An owner employing dominance-based techniques toward their dog who is aggressive toward other dogs can actually cause that dog to not only be more aggressive toward other dogs, due to the added negative association with pain and fear, but also cause the dog to redirect its aggression toward its owner. In which case the problem goes from being something that could possibly be solved via proper training to what is a probable euthanasia case.
Positive reinforcement techniques such as clicker training are gaining momentum, and it's got behaviorists cheering in the streets (or rather, their offices). These techniques avoid the negative associations with pain and fear seen with dominance-based techniques and thus the ramping-up effect on aggression.
Finally, I know that this is a contentious topic and I anticipate anecdotes from those who have used Cesar Millan's or other dominance-based techniques successfully. A few words on that.
First of all, there are always outliers. I saw something recently that I quite liked and determined to borrow that said that between 80-90% of smokers will develop lung cancer, which means that 10-20 out of every 100 smokers will not develop lung cancer. So you will often hear claims such as, "My father smoked two packs a day for forty years and died in his sleep at 85 years old!" And while true, it does not disprove the fact that overall smoking is highly associated with lung cancer.
Also consider that the effect of fear on the cessation of all forms of behavior is fairly well documented. Simply put, a fearful animal will stop doing anything, including what you wanted them to stop doing. A dog that is fearful of inviting a painful punishment can appear to an owner to be "cured" of the unwanted behavior. But in fact, the underlying issue of why this dog was exhibiting the unwanted behavior is still unaddressed. A dog that is fear aggressive toward strangers, for example, is still terrified of strangers but simply stops reacting. Don't confuse this with being a happy, healthy, well-adjusted dog. An animal that has stopping displaying observable fear signals is still fearful, and the use of punishment can contribute to a more unpredictable animal that will give no warning before attacking (AVSAB, 2007)