"an incident should only be included if the bear spray was actually sprayed or if a shot were actually fired."

Well, this doesn't seem right to me. I think the proper evaluation of ANY technology requires the consideration of full context of its use, including educational, legal, economic, societal, and usability aspects.

In general (without regard to the specific articles) it is actually quite scientific not to pre-judge where the problem is before beginning research.

- Suppose bear spray was extremely expensive and canisters were rented with a big deposit against their use?
- Suppose the act of carrying bear spray increased the confidence of people carrying it, and caused them to unwisely enter situations they shouldn't?
- Suppose there were severe laws against unleashing bear spray against a bear unless it was actually attacking?
- Suppose bear spray didn't come with instructions on how to use it?
- Suppose bear spray didn't come with a holster or clip and the only way to carry it was in hand or in a pack?
- Suppose bear spray was so difficult to aim that only an expert could effectively use it?

None of the above have anything to do with the response of a bear to a proper application of spray to its face, but they would be important in an OVERALL evaluation of the technology. And they would also be important in actually figuring out what it is that needs fixing.

I am not suggesting anything about the statistical validity of the articles in question - just that it is too narrow-minded to look at only the end-point (unleashing bear spray or firing a gun) without looking at everything else that may be going on.