Here’s your argument summarized:

When considering the whole sample size of all dogs in a given area, pitbulls are statistically abnormally dangerous because despite being less populous that other races they are responsible for a large amount of the killings caused by dogs.

Is that your argument? Or am I misinterpreting?

Assuming that is your argument, you’re correct in saying that, but what you don’t understand is that “statistically abnormally dangerous” is not the same as dangerous or aggressive. You’re forgetting one of the most important rules in statistics: Correlation does not imply causation. You have a correlation between dog races and violence, and your conclusion is that the race causes the violence, ignoring all other possible explanations for why it could be that there’s a correlation there, for example my example of “some people who mistreat dogs prefer pitbulls, therefore pitbulls are statistically abnormally mistreated”.

Following a couple links from the Wikipedia page on list of fatalities by dogs you will find this quote:

Breed is not an accurate predictor of whether or not a dog will bite.

Which links to this, in which you can find this quote about pitbulls:

controlled studies have not identified this breed group as disproportionately dangerous (…) owners of stigmatized breeds are more likely to have involvement in criminal and/or violent acts—breed correlations may have the owner’s behavior as the underlying causal factor.

Which is very similar to the point I’m trying to make, remember correlation does not imply causation, that is a very slippery slope that anyone with a basic understanding of statistics knows.