I don’t see any reference to controls in the article (and the paper is paywalled, because obviously it would be :D).
That said reading more closely it’s a study of 36 people. 36 how does this get published? And then get an article???
That said reading more closely it’s a study of 36 people. 36 how does this get published? And then get an article???
The article can be summed up with this, "It is possible that humans with higher levels of stress do not have a relaxed relationship with their dogs."
This is what stress does and it's not only applicable to the dogs as I am sure friends, family and co-workers would also have less than a relaxed relationship with the individual.
The authors note that for most of the variables people are talking about, the groups were pretty balanced. Which isn't surprising given they used snowball sampling for their recruitment - that's good at getting homogeneous groups.
Being hard to get a decent sample size isn’t a justification for making unreasonable claims.
A few people have commented on the nature of the samples group (namely that they were homogenous, etc), but that’s not mentioned in the article, and the paper is behind a paywall. If you want a press release for your paper you need to make sure the relevant data is available to people who otherwise only have access to the article.
In one litter you will get such a large range of personalities even with same parents and being raised the same.
And then there’s such a big range of behaviours between the breeds. For example, weimaraners are predisposed to separation anxiety.
In addition, some owners adopt or rescue dogs who had bad experiences.
With the article behind a paywall, I’m guessing a sample size of 70 is way too small to come up with something insightful. How on earth could they come up with a “negative spiral” from the data. Sounds made up.
Anyway, I’ve stopped judging owners based on the behaviour of their dogs.
Having been on the end of that: They don't.
Also, 70 subjects is decently well powered for their analysis, which is actually pretty simple. It's not nearly what I'd prefer, but it's just a bunch of Mann-Whitney U tests and chi-squared tests of association.