According to a report by Merritt Clifton (via Rosalind Arden), pitbulls accounted for 295 of 593 human fatalities due to dogs between 1982-2014, although only making up 6.7% of dogs. But that’s still the second most popular breed, behind only labrador mixes. My observation from walking down the sidewalk is that pitbulls are much more prevalent today in Los Angeles than a half century ago, when they were only vaguely heard of.
In contrast, labradors and lab mixes account for 11.5% of dogs, and only 4 human deaths.
German shepherds, an aggressive/protective breed, are in-between with 15 fatalities and 3.7% of dogs.
Pitbulls, which aren’t particularly big, aren’t the most dangerous dog per capita. The perro de presa canario, a 100+ pound beast, killed 18 people despite being only 0.02% of dogs for sale or adoption. Both are in the molosser class.
Also, wolfish dogs, such as akita, huskies, and wolf-hybrids are pretty scary, as are chows, a wolfish-molosser cross.
Rottweilers are about as dangerous per capita as pit bulls. Dobermans, however, which were notorious when I was a child as WWII guard dogs, have gotten less dangerous: my recollection is that Doberman owners have been breeding for safety while rottweiler owners have been breeding their dogs to be scary.
It would seem like pit bulls could, like dobermans, be bred to keep their good qualities, while breeding out their bad qualities, like a tendency to kill people. But breeding as a hobby or career doesn’t seem to attract as many people with pro-social tendencies as it did in the past, when, for example, Charles Darwin bred pigeons.
It could be that our culture has largely bred out from its more gentlemanly ranks the ruthlessness that helps one be a good breeder, which might explain why dog-breeding today is so much less productive than in the 19th century.
P.S., here is a 2006 Malcolm Gladwell article in The New Yorker on why racial profiling is as bad as profiling pit bulls: “TROUBLEMAKERS: What pit bulls can teach us about profiling.”