Correlation vs causation is irrelevant in most dog bite contexts, where the practical concern is harm avoidance.
If harm avoidance is the goal then why not take it on by promoting better dog upbringing?
Also relevant is that the breed has an effect on the seriousness of injury if an attack does occur. Obviously I'm going to be less fearful of a chihuahua even if they're more aggressive than the terrier. It's the difference between antibiotics and stitches vs. facial reconstruction.
>Or would it be better to only discriminate against dogs and not their owners? We all know how much the dogs use USPS.
I don't know what this means.
>why not take it on by promoting better dog upbringing?
I do promote that. I don't know where you're getting this from.
Back to the idea of harm avoidance:
Let's say we discover that the only factors that determine bite risk are how the dog is raised and how you behave around it. And let's say there is a non-causative correlation showing breed X has a high bite risk. Then it would be reasonable when adopting a puppy to not consider breed X a high bite risk, but it would also be reasonable to consider breed X a high bite risk when encountering one you didn't raise yourself, and therefore e.g. not allow your child to interact with it (or whatever precaution makes one feel comfortable). More information is better.
>I don't know what this means.
I'll explain my meaning with a hypothetical.
What if one breed were responsible for a high number of USPS delivery worker attacks? Promoting this information begs a policy response against this breed. What if most of this breed were owned by a certain demographic? From the outside this looks like a policy response against a demographic with window dressing to justify it. The outcome is what matters.
I agree with your later point. However, dog owners and parents are responsible for their dogs and children (respectively). More information is better, but selecting some specific information here is clearly targeting something or someone, regardless of intent. That is what I take issue with. Target the issue, not the demographic.