I could write a "factual" article claiming hundreds of mass shootings in 2020 (obviously false). I just need to define a "mass shooting" as an incident where four or more people are injured (no deaths required).
Or an equally "factual" article claiming that zero mass shootings in 2020 (also obviously false). I just need to define a "mass shooting" as an incident where twenty or more people are killed.
Exact same dataset, two different and mutually exclusive "facts".
What counts as "injured" or "killed"? If shots are fired and the resulting human stampede kills 4 people, does that count as 4 mass shooting deaths? Obviously the shooter is at fault, but these details affect the interpretation of events.
My undergrad was in statistics. In our capstone course, my professor had us read journal articles and discuss the statistical analyses within. I remember one study we read (peer reviewed, a couple dozen citations), and my professor's take away was, "I can't say it's wrong, but based on the data they gave, I can't for the life of me figure out how they reached their statistical conclusions." So yeah, it's a "fact" that the researchers reached a certain conclusion, but the conclusion itself is not fact.
I don't believe in post-truth, but "Facts are facts; truth is truth" is a philosophical statement, not a practical one. And even then, we have an entire field of philosophy to iron out those details, which we call epistemology.
> What counts as "injured" or "killed"? If shots are fired and the resulting human stampede kills 4 people, does that count as 4 mass shooting deaths? Obviously the shooter is at fault, but these details affect the interpretation of events.
Absolutely!
> I don't believe in post-truth, but "Facts are facts; truth is truth" is a philosophical statement, not a practical one.
In this example, the empirical facts are "hundreds of incidents where four or more people are injured" and "no incidents where twenty or more people are killed". Those facts still exist. The different definitions of "mass shooting" are spin, which obscures facts, but does not eliminate them. Yes, it is hard to pierce the spin to find the facts, but the facts are there somewhere.
You're making a strong statement in favor of fact checking.
People can lie with statistics and people can lie without statistics. The latter is much easier, but the former is possible, as you lay out.
That's why we need to check whether an alleged fact is true, or at least can be confirmed from multiple sources of evidence so it can be accepted as true for the time being. We can also check statistics for anomalies and errors. Statisticians do that all the time.
It is amazingly easy to lie with statistical "facts", through careful sampling, use of technical language, and overly broad or narrow definitions: https://medium.com/@hollymathnerd/how-to-defend-yourself-fro...
I could write a "factual" article claiming hundreds of mass shootings in 2020 (obviously false). I just need to define a "mass shooting" as an incident where four or more people are injured (no deaths required).
Or an equally "factual" article claiming that zero mass shootings in 2020 (also obviously false). I just need to define a "mass shooting" as an incident where twenty or more people are killed.
Exact same dataset, two different and mutually exclusive "facts".