That is true, which is why most of the reports have to have some meta-analysis on them to be useful, but where dog breed and injury type/circumstances are broadly available within the report, breeds like labrador, spaniel, chihuahua, poodle, etc (and other, reasonably recognizable breeds) the injures are almost overwhelmingly related to non-life-threatening injuries and/or unusual circumstances (feral dogs, part of packs, extreme neglect or abuse) while deaths or serious, life-threatening instances where breeds seem reasonably documented, 60%+ are from the three commonly expected breed/types, which very heavily outweighs the percent of those breeds in the population.

If type of dog commonly labeled 'pitbulls' made up 60% of the population and were involved in 60% of attacks, that would basically mean they posed no more threat than any other breed....if they only make up 1% of the population and are involved in 60% of life-threatening attacks, it's fair to say that 'breed' is extremely dangerous. It's much closer to the second example than that first. If you wanted to make a good argument, if you could identify some specific breed that is commonly identified as 'pitbull' but which arguably are 'not' involved in life-threatening attacks, that might be worth highlighting, but unfortunately, just like everyone 'calling everything that looks vaguely like a pitbull, a pitbull'....the instincts that earn then the poor reputation are just as spread out across the group as the physically recognizable traits.

Basically, the response to your comment is 'yeah, but.....' because even though you're right that we probably will never know exactly what breed caused which injury, there is an obvious enough pattern that pretending there isn't a pretty heavy relationship between dogs 'significantly mixed' with pitbull and rottweilers and serious attacks is either intentionally deceiving or ignorant.