Sure it matters, if you don’t normalize for sample size your data is meaningless. Here’s a similar example:
I have a farm that produces both red and green apples, last harvest I had to throw away 100 red apples because they went bad, but only 10 green apples for the same reason.
You would think that red apples spoil 10x easier than green ones, but you’re missing the crucial information that my farm produced 10.000 red apples and 1.000 green ones, so in both cases it represented 1% of the whole amount.
This could be the same thing, if there are more Pitbulls than other races it would be expected that they bite more in absolute numbers.
What? The number of dogs absolutely matters. This is like, basic statistics.
Again, what we’re shown here is the number of fatal injuries, but we don’t necessarily know the distribution of dog breeds. Imagine testing this by putting a child in a room with one of each breed of dog and seeing if they get attacked, except with one breed we put in 200 dogs instead of just one, odds are that yeah, one (or multiple) of those dogs is aggressive because there’s so many of them.
This example is basically what the data above represents, it doesn’t consider that maybe pits actually very rarely bite people, but there’s just so many pits that more bites happen.
It is my personal opinion that pits are a more dangerous breed, but I won’t let that cloud my vision of accurately representing data.
For this to be reflective of an even amount of bites per number of each breed of dog owned, half the dogs owned would have to be pit bulls. That doesn’t match my very casual observations. This site, whose accuracy and validity I haven’t bothered to verify in any way, puts pit bulls at about 20% of the dogs sent to shelters. It seems very likely that the dog fatalities are outsized relative to the number of pit bulls owned in America.
I totally agree with that. The problem is that this graphic doesn’t represent it. The only reason that this graphic looks correct is coincidental. The fact that it’s not an equal #/each is exactly what I want to be shown, but the problem is that this graphic doesn’t show that
You haven’t made any argument against the information presented. You will not get a response from me unless you actually respond instead of repeating the same statement.
Again, you are willfully misunderstanding what the statistic is stating:
Here are the total number of PEOPLE killed by dogs.
Of that number, here’s how many PEOPLE were killed by each breed.
This isn’t tracking bites, or overall attacks, it’s tracking human deaths.
A similar stat would be tracking vehicular accident deaths, if in a year you have accidents involving “brand x” accounting for more vehicular fatalities than all other brands combined, that points to a massive, massive problem with the brand.
It doesn’t matter how many cars there are, that’s not what the stat is tracking.
But if you really want to know, Google says there are around 90 million dogs in the US and 4.5 to 18 million pit bulls depending on how you count. So 5% to 20% of the dog population accounting for 66% of the human deaths.
Then replace every time I said “bite” with “someone died”. That doesn’t change anything about my argument or the validity of it.
In your car example, that is exactly wrong, let’s say that 99 out of 100 cars are Toyota Corollas, and as a whole, they get in 50 fatal accidents every year, but the remaining 1 car is a Ford F150 which got in 1 fatal accident every year. Does this mean that corollas are more dangerous? No! It just means that there are more corollas and therefore more opportunities to kill.
The correct way to represent this is as a percentage of each car brand. 50 accidents divided by 99 corollas is a little less than 50%. 1 accidents divided by 1 F150 is about 100%. According to this, F150’s are actually more dangerous because 100% of them get in fatal accidents, whereas only 50% of corollas get in fatal accidents.
Good on you for looking into the actual data, but the problem is that this specific graphic isn’t showing the actual data. The only reason it looks correct is by coincidence.
The number of dogs doesn’t matter, it’s tracking the number of human beings killed by the breed.
Yeah, Chihuahua might bite you, they aren’t going to kill you.
Sure it matters, if you don’t normalize for sample size your data is meaningless. Here’s a similar example:
You would think that red apples spoil 10x easier than green ones, but you’re missing the crucial information that my farm produced 10.000 red apples and 1.000 green ones, so in both cases it represented 1% of the whole amount.
This could be the same thing, if there are more Pitbulls than other races it would be expected that they bite more in absolute numbers.
What? The number of dogs absolutely matters. This is like, basic statistics.
Again, what we’re shown here is the number of fatal injuries, but we don’t necessarily know the distribution of dog breeds. Imagine testing this by putting a child in a room with one of each breed of dog and seeing if they get attacked, except with one breed we put in 200 dogs instead of just one, odds are that yeah, one (or multiple) of those dogs is aggressive because there’s so many of them.
This example is basically what the data above represents, it doesn’t consider that maybe pits actually very rarely bite people, but there’s just so many pits that more bites happen.
It is my personal opinion that pits are a more dangerous breed, but I won’t let that cloud my vision of accurately representing data.
For this to be reflective of an even amount of bites per number of each breed of dog owned, half the dogs owned would have to be pit bulls. That doesn’t match my very casual observations. This site, whose accuracy and validity I haven’t bothered to verify in any way, puts pit bulls at about 20% of the dogs sent to shelters. It seems very likely that the dog fatalities are outsized relative to the number of pit bulls owned in America.
I totally agree with that. The problem is that this graphic doesn’t represent it. The only reason that this graphic looks correct is coincidental. The fact that it’s not an equal #/each is exactly what I want to be shown, but the problem is that this graphic doesn’t show that
The number of dogs is irrelevant because the statistic is counting human deaths by dogs.
You wouldn’t say “well, how many people are there?” either.
Think about it in the extremes.
If there were a billion poodles out in the world, and they caused 10 human deaths.
And there were 50 terriers out there and they caused 10 human deaths.
Which breed would you buy for your grandmother?
For every five terriers out there, one is killing someone on average. I would go with the Poodle.
You haven’t made any argument against the information presented. You will not get a response from me unless you actually respond instead of repeating the same statement.
Again, you are willfully misunderstanding what the statistic is stating:
Here are the total number of PEOPLE killed by dogs.
Of that number, here’s how many PEOPLE were killed by each breed.
This isn’t tracking bites, or overall attacks, it’s tracking human deaths.
A similar stat would be tracking vehicular accident deaths, if in a year you have accidents involving “brand x” accounting for more vehicular fatalities than all other brands combined, that points to a massive, massive problem with the brand.
It doesn’t matter how many cars there are, that’s not what the stat is tracking.
But if you really want to know, Google says there are around 90 million dogs in the US and 4.5 to 18 million pit bulls depending on how you count. So 5% to 20% of the dog population accounting for 66% of the human deaths.
Then replace every time I said “bite” with “someone died”. That doesn’t change anything about my argument or the validity of it.
In your car example, that is exactly wrong, let’s say that 99 out of 100 cars are Toyota Corollas, and as a whole, they get in 50 fatal accidents every year, but the remaining 1 car is a Ford F150 which got in 1 fatal accident every year. Does this mean that corollas are more dangerous? No! It just means that there are more corollas and therefore more opportunities to kill.
The correct way to represent this is as a percentage of each car brand. 50 accidents divided by 99 corollas is a little less than 50%. 1 accidents divided by 1 F150 is about 100%. According to this, F150’s are actually more dangerous because 100% of them get in fatal accidents, whereas only 50% of corollas get in fatal accidents.
Good on you for looking into the actual data, but the problem is that this specific graphic isn’t showing the actual data. The only reason it looks correct is by coincidence.